HP OpenView GlancePlus · Hewlett-Packard makes no warranty of any kind with regard to this material, including but not limited to, the implied warranties of merchantability and fitness

HP OpenViewGlancePlus

(for HP-UX 11.0 and Beyond)

Dictionary of Performance Metrics

GlancePlus Dictionary of Performance Metrics

ii

Notice

Hewlett-Packard makes no warranty of any kind with regard to thismaterial, including but not limited to, the implied warranties ofmerchantability and fitness for a particular purpose. Hewlett-Packardshall not be held liable for errors contained herein or direct, indirect,special, incidental, or consequential damages in connection with thefurnishing performance, or use of this material.

Warranty. A copy of the specific warranty terms applicable to your Hewlett-Packard product can be obtained from your local Sales and Service Office.Restricted Rights Legend. All rights are reserved. No part of this material may becopied, reproduced, or translated to another language without the prior writtenconsent of Hewlett-Packard Company. The information contained in this material issubject to change without notice.Use, duplication or disclosure by the U.S. Government is subject to restrictions asset forth in subparagraph (c)(1)(ii) of the Rights in Technical Data and ComputerSoftware clause at DFARS 252.227-7013 for DOD agencies, and subparagraphs(c)(1) and (c)(2) of the Commercial Computer Software Restricted Rights clause atFAR 52.227-19 for other agencies.HEWLETT-PACKARD COMPANYUnited States of AmericaCopyright Notices. © Copyright 2002 Hewlett-Packard Company, all rightsreserved. Reproduction, adaptation, or translation of this material without priorwritten permission is prohibited, except as allowed under the copyright laws of theUnited States.Trademark Notices. RISC System/6000 is a trademark and AIX is aregistered trademark of International Business Machines Corporation. Sun andSolaris are registered trademarks of Sun Microsystems, Inc. Adobe ,Acrobat , and PostScript are trademarks of Adobe Systems Incorporated.UNIX is a registered trademark in the United States and other countries,licensed exclusively through X/Open Company Limited. All other product namesare the property of their respective trademark or service mark holders and arehereby acknowledged.


iii

Edition History

This document is not printed. It is distributed on CD with GlancePlusin Adobe Portable Document (.pdf) format and can be viewed onlineand printed as needed. Beginning with Edition 8, part numbers willnot be assigned.

Edition 1 10/99 Accompanies GlancePlus for HP-UX 11.0, releaseC.02.40.00.




Edition 5 03/01 B3691-90056. Accompanies GlancePlus for HP-UX11.0, release C.03.25.00.




Edition 9 06/02 Accompanies GlancePlus for HP-UX 11.0 andbeyond, release C.03.58.00

Conventions

boldface Words in boldface represent the names ofprograms and commands.

computer font Words in computer font represent file names,syntax, directory path names, or text as you shouldenter it on your workstation or terminal, and textthat appears on the screen.

italics Italics are used to emphasize words, phrases, orcharacters in the text, or indicate variables insyntax strings.

Return or Enter keys Depending on your keyboard, one or the otherrepresents the key used to execute a command.


iv

Changes for this Release

This section summarizes the changes made to this document for thisrelease of GlancePlus for HP-UX 11.0 and beyond, release C.03.58.

• The following metric was added:

BYNETIF_INTERVAL_CUM

• The definitions were updated for the following metrics:

BYNETIF_NET_TYPE

GBL_JOBCTL_QUEUE

GBL_NET_IP_FRAGMENTS_RECEIVED

GBL_NET_IP_FWD_DATAGRAMS

GBL_NET_IP_REASSEMBLY_REQUIRED


v

Contents

INTRODUCTION ............................................................................................ 1

METRIC DATA CLASSES ............................................................................... 1

METRIC DEFINITIONS................................................................................ 52

GLOSSARY................................................................................................ 540

GlancePlus Dictionary of Performance MetricsIntroduction

1

IntroductionThis dictionary contains definitions of the GlancePlus performancemetrics on the HP-UX 11.0 and beyond platform. The metric names arelisted first, grouped by data class. The metric definitions are listednext, in alphabetical order. The glossary of performance metric terms isincluded at the end of this document.

The name MeasureWare Agent for UNIX has been replaced with HPOpenView Performance Agent (OV Performance Agent or OVPA)forUNIX and the name PerfView for UNIX has been replaced with OVPerformance Manager for UNIX throughout this documentation.However, the process names and software components operationallyremain MeasureWare Agent (MWA) and PerfView.

Metric Data Classes

Global MetricsGBL_ACTIVE_CPU

GBL_ACTIVE_PROC

GBL_ALIVE_PROC

GBL_BLANK

GBL_CACHE_QUEUE

GBL_CACHE_WAIT_PCT

GBL_CACHE_WAIT_TIME

GBL_CDFS_QUEUE

GBL_CDFS_WAIT_PCT

GBL_CDFS_WAIT_TIME

GBL_COMPLETED_PROC

GBL_CPU_CSWITCH_TIME

GBL_CPU_CSWITCH_TIME_CUM

GBL_CPU_CSWITCH_UTIL

GBL_CPU_CSWITCH_UTIL_CUM

GBL_CPU_CSWITCH_UTIL_HIGH

GBL_CPU_IDLE_TIME

GBL_CPU_IDLE_TIME_CUM

GBL_CPU_IDLE_UTIL

GBL_CPU_IDLE_UTIL_CUM

N O T E :


2

GBL_CPU_IDLE_UTIL_HIGH

GBL_CPU_INTERRUPT_TIME

GBL_CPU_INTERRUPT_TIME_CUM

GBL_CPU_INTERRUPT_UTIL

GBL_CPU_INTERRUPT_UTIL_CUM

GBL_CPU_INTERRUPT_UTIL_HIGH

GBL_CPU_NICE_TIME

GBL_CPU_NICE_TIME_CUM

GBL_CPU_NICE_UTIL

GBL_CPU_NICE_UTIL_CUM

GBL_CPU_NICE_UTIL_HIGH

GBL_CPU_NNICE_TIME

GBL_CPU_NNICE_TIME_CUM

GBL_CPU_NNICE_UTIL

GBL_CPU_NNICE_UTIL_CUM

GBL_CPU_NNICE_UTIL_HIGH

GBL_CPU_NORMAL_TIME

GBL_CPU_NORMAL_TIME_CUM

GBL_CPU_NORMAL_UTIL

GBL_CPU_NORMAL_UTIL_CUM

GBL_CPU_NORMAL_UTIL_HIGH

GBL_CPU_QUEUE

GBL_CPU_REALTIME_TIME

GBL_CPU_REALTIME_TIME_CUM

GBL_CPU_REALTIME_UTIL

GBL_CPU_REALTIME_UTIL_CUM

GBL_CPU_REALTIME_UTIL_HIGH

GBL_CPU_SYSCALL_TIME

GBL_CPU_SYSCALL_TIME_CUM

GBL_CPU_SYSCALL_UTIL

GBL_CPU_SYSCALL_UTIL_CUM

GBL_CPU_SYSCALL_UTIL_HIGH

GBL_CPU_SYS_MODE_TIME

GBL_CPU_SYS_MODE_TIME_CUM

GBL_CPU_SYS_MODE_UTIL

GBL_CPU_SYS_MODE_UTIL_CUM

GlancePlus Dictionary of Performance MetricsMetric Data Classes

3

GBL_CPU_TOTAL_TIME

GBL_CPU_TOTAL_TIME_CUM

GBL_CPU_TOTAL_UTIL

GBL_CPU_TOTAL_UTIL_CUM

GBL_CPU_TOTAL_UTIL_HIGH

GBL_CPU_TRAP_TIME

GBL_CPU_TRAP_TIME_CUM

GBL_CPU_TRAP_UTIL

GBL_CPU_TRAP_UTIL_CUM

GBL_CPU_TRAP_UTIL_HIGH

GBL_CPU_USER_MODE_TIME

GBL_CPU_USER_MODE_TIME_CUM

GBL_CPU_USER_MODE_UTIL

GBL_CPU_USER_MODE_UTIL_CUM

GBL_CPU_VFAULT_TIME

GBL_CPU_VFAULT_TIME_CUM

GBL_CPU_VFAULT_UTIL

GBL_CPU_VFAULT_UTIL_CUM

GBL_CPU_VFAULT_UTIL_HIGH

GBL_CSWITCH_RATE

GBL_CSWITCH_RATE_CUM

GBL_CSWITCH_RATE_HIGH

GBL_DISK_FS_BYTE

GBL_DISK_FS_BYTE_CUM

GBL_DISK_FS_IO

GBL_DISK_FS_IO_CUM

GBL_DISK_FS_IO_PCT

GBL_DISK_FS_IO_PCT_CUM

GBL_DISK_FS_IO_RATE

GBL_DISK_FS_IO_RATE_CUM

GBL_DISK_FS_READ

GBL_DISK_FS_READ_RATE

GBL_DISK_FS_WRITE

GBL_DISK_FS_WRITE_RATE

GBL_DISK_LOGL_BYTE_RATE

GBL_DISK_LOGL_IO


4

GBL_DISK_LOGL_IO_CUM

GBL_DISK_LOGL_IO_RATE

GBL_DISK_LOGL_IO_RATE_CUM

GBL_DISK_LOGL_READ

GBL_DISK_LOGL_READ_BYTE

GBL_DISK_LOGL_READ_BYTE_CUM

GBL_DISK_LOGL_READ_BYTE_RATE

GBL_DISK_LOGL_READ_CUM

GBL_DISK_LOGL_READ_PCT

GBL_DISK_LOGL_READ_PCT_CUM

GBL_DISK_LOGL_READ_RATE

GBL_DISK_LOGL_READ_RATE_CUM

GBL_DISK_LOGL_WRITE

GBL_DISK_LOGL_WRITE_BYTE

GBL_DISK_LOGL_WRITE_BYTE_CUM

GBL_DISK_LOGL_WRITE_BYTE_RATE

GBL_DISK_LOGL_WRITE_CUM

GBL_DISK_LOGL_WRITE_PCT

GBL_DISK_LOGL_WRITE_PCT_CUM

GBL_DISK_LOGL_WRITE_RATE

GBL_DISK_LOGL_WRITE_RATE_CUM

GBL_DISK_PHYS_BYTE

GBL_DISK_PHYS_BYTE_RATE

GBL_DISK_PHYS_IO

GBL_DISK_PHYS_IO_CUM

GBL_DISK_PHYS_IO_RATE

GBL_DISK_PHYS_IO_RATE_CUM

GBL_DISK_PHYS_READ

GBL_DISK_PHYS_READ_BYTE

GBL_DISK_PHYS_READ_BYTE_CUM

GBL_DISK_PHYS_READ_BYTE_RATE

GBL_DISK_PHYS_READ_CUM

GBL_DISK_PHYS_READ_PCT

GBL_DISK_PHYS_READ_PCT_CUM

GBL_DISK_PHYS_READ_RATE

GBL_DISK_PHYS_READ_RATE_CUM


5

GBL_DISK_PHYS_WRITE

GBL_DISK_PHYS_WRITE_BYTE

GBL_DISK_PHYS_WRITE_BYTE_CUM

GBL_DISK_PHYS_WRITE_BYTE_RATE

GBL_DISK_PHYS_WRITE_CUM

GBL_DISK_PHYS_WRITE_PCT

GBL_DISK_PHYS_WRITE_PCT_CUM

GBL_DISK_PHYS_WRITE_RATE

GBL_DISK_PHYS_WRITE_RATE_CUM

GBL_DISK_QUEUE

GBL_DISK_RAW_BYTE

GBL_DISK_RAW_BYTE_CUM

GBL_DISK_RAW_IO

GBL_DISK_RAW_IO_CUM

GBL_DISK_RAW_IO_PCT

GBL_DISK_RAW_IO_PCT_CUM

GBL_DISK_RAW_IO_RATE

GBL_DISK_RAW_IO_RATE_CUM

GBL_DISK_RAW_READ

GBL_DISK_RAW_READ_RATE

GBL_DISK_RAW_WRITE

GBL_DISK_RAW_WRITE_RATE

GBL_DISK_REM_FS_BYTE

GBL_DISK_REM_FS_BYTE_CUM

GBL_DISK_REM_FS_IO

GBL_DISK_REM_FS_IO_CUM

GBL_DISK_REM_FS_IO_PCT

GBL_DISK_REM_FS_IO_PCT_CUM

GBL_DISK_REM_FS_IO_RATE

GBL_DISK_REM_FS_IO_RATE_CUM

GBL_DISK_REM_LOGL_READ

GBL_DISK_REM_LOGL_READ_BYTE

GBL_DISK_REM_LOGL_READ_BYTE_CUM

GBL_DISK_REM_LOGL_READ_CUM

GBL_DISK_REM_LOGL_READ_PCT

GBL_DISK_REM_LOGL_READ_PCT_CUM


6

GBL_DISK_REM_LOGL_READ_RATE

GBL_DISK_REM_LOGL_READ_RATE_CUM

GBL_DISK_REM_LOGL_WRITE

GBL_DISK_REM_LOGL_WRITE_BYTE

GBL_DISK_REM_LOGL_WRITE_BYTE_CUM

GBL_DISK_REM_LOGL_WRITE_CUM

GBL_DISK_REM_LOGL_WRITE_PCT

GBL_DISK_REM_LOGL_WRITE_PCT_CUM

GBL_DISK_REM_LOGL_WRITE_RATE

GBL_DISK_REM_LOGL_WRITE_RATE_CUM

GBL_DISK_REM_PHYS_READ

GBL_DISK_REM_PHYS_READ_BYTE

GBL_DISK_REM_PHYS_READ_BYTE_CUM

GBL_DISK_REM_PHYS_READ_CUM

GBL_DISK_REM_PHYS_READ_PCT

GBL_DISK_REM_PHYS_READ_PCT_CUM

GBL_DISK_REM_PHYS_READ_RATE

GBL_DISK_REM_PHYS_READ_RATE_CUM

GBL_DISK_REM_PHYS_WRITE

GBL_DISK_REM_PHYS_WRITE_BYTE

GBL_DISK_REM_PHYS_WRITE_BYTE_CUM

GBL_DISK_REM_PHYS_WRITE_CUM

GBL_DISK_REM_PHYS_WRITE_PCT

GBL_DISK_REM_PHYS_WRITE_PCT_CUM

GBL_DISK_REM_PHYS_WRITE_RATE

GBL_DISK_REM_PHYS_WRITE_RATE_CUM

GBL_DISK_REM_RAW_BYTE

GBL_DISK_REM_RAW_BYTE_CUM

GBL_DISK_REM_RAW_IO

GBL_DISK_REM_RAW_IO_CUM

GBL_DISK_REM_RAW_IO_PCT

GBL_DISK_REM_RAW_IO_PCT_CUM

GBL_DISK_REM_RAW_IO_RATE

GBL_DISK_REM_RAW_IO_RATE_CUM

GBL_DISK_REM_SYSTEM_BYTE

GBL_DISK_REM_SYSTEM_BYTE_CUM


7

GBL_DISK_REM_SYSTEM_IO

GBL_DISK_REM_SYSTEM_IO_CUM

GBL_DISK_REM_SYSTEM_IO_PCT

GBL_DISK_REM_SYSTEM_IO_PCT_CUM

GBL_DISK_REM_SYSTEM_IO_RATE

GBL_DISK_REM_SYSTEM_IO_RATE_CUM

GBL_DISK_REM_VM_BYTE

GBL_DISK_REM_VM_BYTE_CUM

GBL_DISK_REM_VM_IO

GBL_DISK_REM_VM_IO_CUM

GBL_DISK_REM_VM_IO_PCT

GBL_DISK_REM_VM_IO_PCT_CUM

GBL_DISK_REM_VM_IO_RATE

GBL_DISK_REM_VM_IO_RATE_CUM

GBL_DISK_SUBSYSTEM_QUEUE

GBL_DISK_SUBSYSTEM_WAIT_PCT

GBL_DISK_SYSTEM_BYTE

GBL_DISK_SYSTEM_BYTE_CUM

GBL_DISK_SYSTEM_IO

GBL_DISK_SYSTEM_IO_CUM

GBL_DISK_SYSTEM_IO_PCT

GBL_DISK_SYSTEM_IO_PCT_CUM

GBL_DISK_SYSTEM_IO_RATE

GBL_DISK_SYSTEM_IO_RATE_CUM

GBL_DISK_SYSTEM_READ

GBL_DISK_SYSTEM_READ_RATE

GBL_DISK_SYSTEM_WRITE

GBL_DISK_SYSTEM_WRITE_RATE

GBL_DISK_TIME_PEAK

GBL_DISK_UTIL

GBL_DISK_UTIL_PEAK

GBL_DISK_UTIL_PEAK_CUM

GBL_DISK_UTIL_PEAK_HIGH

GBL_DISK_UTIL_PEAK_OTHERS

GBL_DISK_UTIL_PEAK_VM

GBL_DISK_VM_BYTE


8

GBL_DISK_VM_BYTE_CUM

GBL_DISK_VM_IO

GBL_DISK_VM_IO_CUM

GBL_DISK_VM_IO_PCT

GBL_DISK_VM_IO_PCT_CUM

GBL_DISK_VM_IO_RATE

GBL_DISK_VM_IO_RATE_CUM

GBL_DISK_VM_READ

GBL_DISK_VM_READ_CUM

GBL_DISK_VM_READ_RATE

GBL_DISK_VM_READ_RATE_CUM

GBL_DISK_VM_READ_RATE_HIGH

GBL_DISK_VM_WRITE

GBL_DISK_VM_WRITE_CUM

GBL_DISK_VM_WRITE_RATE

GBL_DISK_VM_WRITE_RATE_CUM

GBL_DISK_VM_WRITE_RATE_HIGH

GBL_DISK_WAIT_PCT

GBL_DISK_WAIT_TIME

GBL_FS_SPACE_UTIL_PEAK

GBL_GRAPHICS_QUEUE

GBL_GRAPHICS_WAIT_PCT

GBL_GRAPHICS_WAIT_TIME

GBL_INODE_QUEUE

GBL_INODE_WAIT_PCT

GBL_INODE_WAIT_TIME

GBL_INTERRUPT_RATE

GBL_INTERRUPT_RATE_CUM

GBL_INTERRUPT_RATE_HIGH

GBL_INTERVAL

GBL_INTERVAL_CUM

GBL_IPC_QUEUE

GBL_IPC_SUBSYSTEM_QUEUE

GBL_IPC_SUBSYSTEM_WAIT_PCT

GBL_IPC_WAIT_PCT

GBL_IPC_WAIT_TIME


9

GBL_JOBCTL_QUEUE

GBL_JOBCTL_WAIT_PCT

GBL_JOBCTL_WAIT_TIME

GBL_LAN_QUEUE

GBL_LAN_WAIT_PCT

GBL_LAN_WAIT_TIME

GBL_LOST_MI_TRACE_BUFFERS

GBL_MACHINE

GBL_MACHINE_MODEL

GBL_MEM_ACTIVE_VIRT

GBL_MEM_ACTIVE_VIRT_UTIL

GBL_MEM_AVAIL

GBL_MEM_CACHE

GBL_MEM_CACHE_HIT

GBL_MEM_CACHE_HIT_CUM

GBL_MEM_CACHE_HIT_PCT

GBL_MEM_CACHE_HIT_PCT_CUM

GBL_MEM_CACHE_HIT_PCT_HIGH

GBL_MEM_CACHE_UTIL

GBL_MEM_CACHE_WRITE_HIT

GBL_MEM_CACHE_WRITE_HIT_CUM

GBL_MEM_CACHE_WRITE_HIT_PCT

GBL_MEM_CACHE_WRITE_HIT_PCT_CUM

GBL_MEM_DNLC_HIT

GBL_MEM_DNLC_HIT_CUM

GBL_MEM_DNLC_HIT_PCT

GBL_MEM_DNLC_HIT_PCT_CUM

GBL_MEM_DNLC_HIT_PCT_HIGH

GBL_MEM_DNLC_LONGS

GBL_MEM_DNLC_LONGS_CUM

GBL_MEM_DNLC_LONGS_PCT

GBL_MEM_DNLC_LONGS_PCT_CUM

GBL_MEM_DNLC_LONGS_PCT_HIGH

GBL_MEM_FREE

GBL_MEM_FREE_UTIL

GBL_MEM_PAGEIN


10

GBL_MEM_PAGEIN_BYTE

GBL_MEM_PAGEIN_BYTE_CUM

GBL_MEM_PAGEIN_BYTE_RATE

GBL_MEM_PAGEIN_BYTE_RATE_CUM

GBL_MEM_PAGEIN_BYTE_RATE_HIGH

GBL_MEM_PAGEIN_CUM

GBL_MEM_PAGEIN_RATE

GBL_MEM_PAGEIN_RATE_CUM

GBL_MEM_PAGEIN_RATE_HIGH

GBL_MEM_PAGEOUT

GBL_MEM_PAGEOUT_BYTE

GBL_MEM_PAGEOUT_BYTE_CUM

GBL_MEM_PAGEOUT_BYTE_RATE

GBL_MEM_PAGEOUT_BYTE_RATE_CUM

GBL_MEM_PAGEOUT_BYTE_RATE_HIGH

GBL_MEM_PAGEOUT_CUM

GBL_MEM_PAGEOUT_RATE

GBL_MEM_PAGEOUT_RATE_CUM

GBL_MEM_PAGEOUT_RATE_HIGH

GBL_MEM_PAGE_FAULT

GBL_MEM_PAGE_FAULT_CUM

GBL_MEM_PAGE_FAULT_RATE

GBL_MEM_PAGE_FAULT_RATE_CUM

GBL_MEM_PAGE_FAULT_RATE_HIGH

GBL_MEM_PAGE_REQUEST

GBL_MEM_PAGE_REQUEST_CUM

GBL_MEM_PAGE_REQUEST_RATE

GBL_MEM_PAGE_REQUEST_RATE_CUM

GBL_MEM_PAGE_REQUEST_RATE_HIGH

GBL_MEM_PAGE_SIZE_MAX

GBL_MEM_PHYS

GBL_MEM_QUEUE

GBL_MEM_SWAP

GBL_MEM_SWAPIN

GBL_MEM_SWAPIN_BYTE

GBL_MEM_SWAPIN_BYTE_CUM


11

GBL_MEM_SWAPIN_BYTE_RATE

GBL_MEM_SWAPIN_BYTE_RATE_CUM

GBL_MEM_SWAPIN_BYTE_RATE_HIGH

GBL_MEM_SWAPIN_CUM

GBL_MEM_SWAPIN_RATE

GBL_MEM_SWAPIN_RATE_CUM

GBL_MEM_SWAPIN_RATE_HIGH

GBL_MEM_SWAPOUT

GBL_MEM_SWAPOUT_BYTE

GBL_MEM_SWAPOUT_BYTE_CUM

GBL_MEM_SWAPOUT_BYTE_RATE

GBL_MEM_SWAPOUT_BYTE_RATE_CUM

GBL_MEM_SWAPOUT_BYTE_RATE_HIGH

GBL_MEM_SWAPOUT_CUM

GBL_MEM_SWAPOUT_RATE

GBL_MEM_SWAPOUT_RATE_CUM

GBL_MEM_SWAPOUT_RATE_HIGH

GBL_MEM_SWAP_1_MIN_RATE

GBL_MEM_SWAP_CUM

GBL_MEM_SWAP_RATE

GBL_MEM_SWAP_RATE_CUM

GBL_MEM_SWAP_RATE_HIGH

GBL_MEM_SYS

GBL_MEM_SYS_AND_CACHE_UTIL

GBL_MEM_SYS_UTIL

GBL_MEM_USER

GBL_MEM_USER_UTIL

GBL_MEM_UTIL

GBL_MEM_UTIL_CUM

GBL_MEM_UTIL_HIGH

GBL_MEM_VIRT

GBL_MEM_WAIT_PCT

GBL_MEM_WAIT_TIME

GBL_MSG_QUEUE

GBL_MSG_WAIT_PCT

GBL_MSG_WAIT_TIME


12

GBL_NETWORK_SUBSYSTEM_QUEUE

GBL_NETWORK_SUBSYSTEM_WAIT_PCT

GBL_NET_COLLISION

GBL_NET_COLLISION_1_MIN_RATE

GBL_NET_COLLISION_CUM

GBL_NET_COLLISION_PCT

GBL_NET_COLLISION_PCT_CUM

GBL_NET_COLLISION_RATE

GBL_NET_DEFERRED

GBL_NET_DEFERRED_CUM

GBL_NET_DEFERRED_PCT

GBL_NET_DEFERRED_PCT_CUM

GBL_NET_DEFERRED_RATE

GBL_NET_DEFERRED_RATE_CUM

GBL_NET_ERROR

GBL_NET_ERROR_1_MIN_RATE

GBL_NET_ERROR_CUM

GBL_NET_ERROR_RATE

GBL_NET_IN_ERROR

GBL_NET_IN_ERROR_CUM

GBL_NET_IN_ERROR_PCT

GBL_NET_IN_ERROR_PCT_CUM

GBL_NET_IN_ERROR_RATE

GBL_NET_IN_ERROR_RATE_CUM

GBL_NET_IN_PACKET

GBL_NET_IN_PACKET_CUM

GBL_NET_IN_PACKET_RATE

GBL_NET_IP_FRAGMENTS_RECEIVED

GBL_NET_IP_FWD_DATAGRAMS

GBL_NET_IP_REASSEMBLY_REQUIRED

GBL_NET_OUTQUEUE

GBL_NET_OUT_ERROR

GBL_NET_OUT_ERROR_CUM

GBL_NET_OUT_ERROR_PCT

GBL_NET_OUT_ERROR_PCT_CUM

GBL_NET_OUT_ERROR_RATE


13

GBL_NET_OUT_ERROR_RATE_CUM

GBL_NET_OUT_PACKET

GBL_NET_OUT_PACKET_CUM

GBL_NET_OUT_PACKET_RATE

GBL_NET_PACKET

GBL_NET_PACKET_RATE

GBL_NFS_CALL

GBL_NFS_CALL_RATE

GBL_NFS_CLIENT_BAD_CALL

GBL_NFS_CLIENT_BAD_CALL_CUM

GBL_NFS_CLIENT_BIOD

GBL_NFS_CLIENT_BYTE

GBL_NFS_CLIENT_BYTE_CUM

GBL_NFS_CLIENT_CALL

GBL_NFS_CLIENT_CALL_CUM

GBL_NFS_CLIENT_CALL_RATE

GBL_NFS_CLIENT_IDLE_BIOD

GBL_NFS_CLIENT_IO

GBL_NFS_CLIENT_IO_CUM

GBL_NFS_CLIENT_IO_PCT

GBL_NFS_CLIENT_IO_PCT_CUM

GBL_NFS_CLIENT_IO_RATE

GBL_NFS_CLIENT_IO_RATE_CUM

GBL_NFS_CLIENT_PHYS_TIME

GBL_NFS_CLIENT_PHYS_TIME_CUM

GBL_NFS_CLIENT_READ_BYTE_RATE

GBL_NFS_CLIENT_READ_BYTE_RATE_CUM

GBL_NFS_CLIENT_READ_RATE

GBL_NFS_CLIENT_READ_RATE_CUM

GBL_NFS_CLIENT_SERVICE_QUEUE

GBL_NFS_CLIENT_SERVICE_QUEUE_CUM

GBL_NFS_CLIENT_SERVICE_TIME

GBL_NFS_CLIENT_SERVICE_TIME_CUM

GBL_NFS_CLIENT_WRITE_BYTE_RATE

GBL_NFS_CLIENT_WRITE_BYTE_RATE_CUM

GBL_NFS_CLIENT_WRITE_RATE


14

GBL_NFS_CLIENT_WRITE_RATE_CUM

GBL_NFS_LOGL_READ

GBL_NFS_LOGL_READ_BYTE

GBL_NFS_LOGL_READ_BYTE_CUM

GBL_NFS_LOGL_READ_CUM

GBL_NFS_LOGL_READ_PCT

GBL_NFS_LOGL_READ_PCT_CUM

GBL_NFS_LOGL_READ_RATE

GBL_NFS_LOGL_READ_RATE_CUM

GBL_NFS_LOGL_WRITE

GBL_NFS_LOGL_WRITE_BYTE

GBL_NFS_LOGL_WRITE_BYTE_CUM

GBL_NFS_LOGL_WRITE_CUM

GBL_NFS_LOGL_WRITE_PCT

GBL_NFS_LOGL_WRITE_PCT_CUM

GBL_NFS_LOGL_WRITE_RATE

GBL_NFS_LOGL_WRITE_RATE_CUM

GBL_NFS_QUEUE

GBL_NFS_SERVER_BAD_CALL

GBL_NFS_SERVER_BAD_CALL_CUM

GBL_NFS_SERVER_BYTE

GBL_NFS_SERVER_BYTE_CUM

GBL_NFS_SERVER_CALL

GBL_NFS_SERVER_CALL_CUM

GBL_NFS_SERVER_CALL_RATE

GBL_NFS_SERVER_IO

GBL_NFS_SERVER_IO_CUM

GBL_NFS_SERVER_IO_PCT

GBL_NFS_SERVER_IO_PCT_CUM

GBL_NFS_SERVER_IO_RATE

GBL_NFS_SERVER_IO_RATE_CUM

GBL_NFS_SERVER_READ_BYTE_RATE

GBL_NFS_SERVER_READ_BYTE_RATE_CUM

GBL_NFS_SERVER_READ_RATE

GBL_NFS_SERVER_READ_RATE_CUM

GBL_NFS_SERVER_SERVICE_TIME


15

GBL_NFS_SERVER_SERVICE_TIME_CUM

GBL_NFS_SERVER_WRITE_BYTE_RATE

GBL_NFS_SERVER_WRITE_BYTE_RATE_CUM

GBL_NFS_SERVER_WRITE_RATE

GBL_NFS_SERVER_WRITE_RATE_CUM

GBL_NFS_WAIT_PCT

GBL_NFS_WAIT_TIME

GBL_NODENAME

GBL_NUM_APP

GBL_NUM_APP_PRM

GBL_NUM_CPU

GBL_NUM_DISK

GBL_NUM_NETWORK

GBL_NUM_SWAP

GBL_NUM_TT

GBL_NUM_USER

GBL_NUM_VG

GBL_OSKERNELTYPE

GBL_OSKERNELTYPE_INT

GBL_OSNAME

GBL_OSRELEASE

GBL_OSVERSION

GBL_OTHER_IO_QUEUE

GBL_OTHER_IO_WAIT_PCT

GBL_OTHER_IO_WAIT_TIME

GBL_OTHER_QUEUE

GBL_OTHER_WAIT_PCT

GBL_OTHER_WAIT_TIME

GBL_PIPE_QUEUE

GBL_PIPE_WAIT_PCT

GBL_PIPE_WAIT_TIME

GBL_PRI_QUEUE

GBL_PRI_WAIT_PCT

GBL_PRI_WAIT_TIME

GBL_PRM_MEM_UTIL

GBL_PROC_RUN_TIME


16

GBL_PROC_SAMPLE

GBL_RPC_QUEUE

GBL_RPC_WAIT_PCT

GBL_RPC_WAIT_TIME

GBL_RUN_QUEUE

GBL_RUN_QUEUE_CUM

GBL_RUN_QUEUE_HIGH

GBL_SAMPLE

GBL_SEM_QUEUE

GBL_SEM_WAIT_PCT

GBL_SEM_WAIT_TIME

GBL_SERIALNO

GBL_SLEEP_QUEUE

GBL_SLEEP_WAIT_PCT

GBL_SLEEP_WAIT_TIME

GBL_SOCKET_QUEUE

GBL_SOCKET_WAIT_PCT

GBL_SOCKET_WAIT_TIME

GBL_STARTDATE

GBL_STARTED_PROC

GBL_STARTTIME

GBL_STATDATE

GBL_STATTIME

GBL_STREAM_QUEUE

GBL_STREAM_WAIT_PCT

GBL_STREAM_WAIT_TIME

GBL_SWAP_RESERVED_ONLY_UTIL

GBL_SWAP_SPACE_AVAIL

GBL_SWAP_SPACE_AVAIL_KB

GBL_SWAP_SPACE_DEVICE_UTIL

GBL_SWAP_SPACE_FS_UTIL

GBL_SWAP_SPACE_RESERVED

GBL_SWAP_SPACE_RESERVED_UTIL

GBL_SWAP_SPACE_USED

GBL_SWAP_SPACE_USED_UTIL

GBL_SWAP_SPACE_UTIL


17

GBL_SWAP_SPACE_UTIL_CUM

GBL_SWAP_SPACE_UTIL_HIGH

GBL_SYSCALL_RATE

GBL_SYSCALL_RATE_CUM

GBL_SYSCALL_RATE_HIGH

GBL_SYSTEM_ID

GBL_SYSTEM_TYPE

GBL_SYSTEM_UPTIME_HOURS

GBL_SYS_QUEUE

GBL_SYS_WAIT_PCT

GBL_SYS_WAIT_TIME

GBL_TERM_IO_QUEUE

GBL_TERM_IO_WAIT_PCT

GBL_TERM_IO_WAIT_TIME

GBL_TT_OVERFLOW_COUNT

Table MetricsTBL_BUFFER_CACHE_AVAIL

TBL_BUFFER_CACHE_HIGH

TBL_BUFFER_CACHE_MAX

TBL_BUFFER_CACHE_MIN

TBL_BUFFER_CACHE_USED

TBL_BUFFER_HEADER_AVAIL

TBL_BUFFER_HEADER_USED

TBL_BUFFER_HEADER_UTIL

TBL_BUFFER_HEADER_UTIL_HIGH

TBL_DNLC_CACHE_AVAIL

TBL_FILE_LOCK_AVAIL

TBL_FILE_LOCK_USED

TBL_FILE_LOCK_UTIL

TBL_FILE_LOCK_UTIL_HIGH

TBL_FILE_TABLE_AVAIL

TBL_FILE_TABLE_USED

TBL_FILE_TABLE_UTIL

TBL_FILE_TABLE_UTIL_HIGH

TBL_INODE_CACHE_AVAIL

TBL_INODE_CACHE_HIGH


18

TBL_INODE_CACHE_USED

TBL_MSG_BUFFER_AVAIL

TBL_MSG_BUFFER_HIGH

TBL_MSG_BUFFER_USED

TBL_MSG_TABLE_AVAIL

TBL_MSG_TABLE_USED

TBL_MSG_TABLE_UTIL

TBL_MSG_TABLE_UTIL_HIGH

TBL_PROC_TABLE_AVAIL

TBL_PROC_TABLE_USED

TBL_PROC_TABLE_UTIL

TBL_PROC_TABLE_UTIL_HIGH

TBL_PTY_AVAIL

TBL_PTY_USED

TBL_PTY_UTIL

TBL_PTY_UTIL_HIGH

TBL_SEM_TABLE_AVAIL

TBL_SEM_TABLE_USED

TBL_SEM_TABLE_UTIL

TBL_SEM_TABLE_UTIL_HIGH

TBL_SHMEM_AVAIL

TBL_SHMEM_REQUESTED

TBL_SHMEM_TABLE_AVAIL

TBL_SHMEM_TABLE_USED

TBL_SHMEM_TABLE_UTIL

TBL_SHMEM_TABLE_UTIL_HIGH

Process MetricsPROC_ACTIVE_PROC

PROC_APP_ID

PROC_APP_NAME

PROC_CACHE_WAIT_PCT

PROC_CACHE_WAIT_PCT_CUM

PROC_CACHE_WAIT_TIME

PROC_CACHE_WAIT_TIME_CUM

PROC_CDFS_WAIT_PCT

PROC_CDFS_WAIT_PCT_CUM


19

PROC_CDFS_WAIT_TIME

PROC_CDFS_WAIT_TIME_CUM

PROC_CLOSE

PROC_CLOSE_CUM

PROC_CPU_CSWITCH_TIME

PROC_CPU_CSWITCH_TIME_CUM

PROC_CPU_CSWITCH_UTIL

PROC_CPU_CSWITCH_UTIL_CUM

PROC_CPU_INTERRUPT_TIME

PROC_CPU_INTERRUPT_TIME_CUM

PROC_CPU_INTERRUPT_UTIL

PROC_CPU_INTERRUPT_UTIL_CUM

PROC_CPU_LAST_USED

PROC_CPU_NICE_TIME

PROC_CPU_NICE_TIME_CUM

PROC_CPU_NICE_UTIL

PROC_CPU_NICE_UTIL_CUM

PROC_CPU_NNICE_TIME

PROC_CPU_NNICE_TIME_CUM

PROC_CPU_NNICE_UTIL

PROC_CPU_NNICE_UTIL_CUM

PROC_CPU_NORMAL_TIME

PROC_CPU_NORMAL_TIME_CUM

PROC_CPU_NORMAL_UTIL

PROC_CPU_NORMAL_UTIL_CUM

PROC_CPU_REALTIME_TIME

PROC_CPU_REALTIME_TIME_CUM

PROC_CPU_REALTIME_UTIL

PROC_CPU_REALTIME_UTIL_CUM

PROC_CPU_SWITCHES

PROC_CPU_SWITCHES_CUM

PROC_CPU_SYSCALL_TIME

PROC_CPU_SYSCALL_TIME_CUM

PROC_CPU_SYSCALL_UTIL

PROC_CPU_SYSCALL_UTIL_CUM

PROC_CPU_SYS_MODE_TIME


20

PROC_CPU_SYS_MODE_TIME_CUM

PROC_CPU_SYS_MODE_UTIL

PROC_CPU_SYS_MODE_UTIL_CUM

PROC_CPU_TOTAL_TIME

PROC_CPU_TOTAL_TIME_CUM

PROC_CPU_TOTAL_UTIL

PROC_CPU_TOTAL_UTIL_CUM

PROC_CPU_TRAP_COUNT

PROC_CPU_TRAP_COUNT_CUM

PROC_CPU_USER_MODE_TIME

PROC_CPU_USER_MODE_TIME_CUM

PROC_CPU_USER_MODE_UTIL

PROC_CPU_USER_MODE_UTIL_CUM

PROC_DISK_FS_READ

PROC_DISK_FS_READ_CUM

PROC_DISK_FS_READ_RATE

PROC_DISK_FS_WRITE

PROC_DISK_FS_WRITE_CUM

PROC_DISK_FS_WRITE_RATE

PROC_DISK_LOGL_IO

PROC_DISK_LOGL_IO_CUM

PROC_DISK_LOGL_IO_RATE

PROC_DISK_LOGL_IO_RATE_CUM

PROC_DISK_LOGL_READ

PROC_DISK_LOGL_READ_CUM

PROC_DISK_LOGL_READ_RATE

PROC_DISK_LOGL_WRITE

PROC_DISK_LOGL_WRITE_CUM

PROC_DISK_LOGL_WRITE_RATE

PROC_DISK_PHYS_IO_RATE

PROC_DISK_PHYS_IO_RATE_CUM

PROC_DISK_PHYS_READ

PROC_DISK_PHYS_READ_CUM

PROC_DISK_PHYS_READ_RATE

PROC_DISK_PHYS_WRITE

PROC_DISK_PHYS_WRITE_CUM


21

PROC_DISK_PHYS_WRITE_RATE

PROC_DISK_RAW_READ

PROC_DISK_RAW_READ_CUM

PROC_DISK_RAW_READ_RATE

PROC_DISK_RAW_WRITE

PROC_DISK_RAW_WRITE_CUM

PROC_DISK_RAW_WRITE_RATE

PROC_DISK_REM_LOGL_READ

PROC_DISK_REM_LOGL_READ_CUM

PROC_DISK_REM_LOGL_READ_RATE

PROC_DISK_REM_LOGL_WRITE

PROC_DISK_REM_LOGL_WRITE_CUM

PROC_DISK_REM_LOGL_WRITE_RATE

PROC_DISK_REM_PHYS_READ

PROC_DISK_REM_PHYS_READ_CUM

PROC_DISK_REM_PHYS_READ_RATE

PROC_DISK_REM_PHYS_WRITE

PROC_DISK_REM_PHYS_WRITE_CUM

PROC_DISK_REM_PHYS_WRITE_RATE

PROC_DISK_SUBSYSTEM_WAIT_PCT

PROC_DISK_SUBSYSTEM_WAIT_PCT_CUM

PROC_DISK_SUBSYSTEM_WAIT_TIME

PROC_DISK_SUBSYSTEM_WAIT_TIME_CUM

PROC_DISK_SYSTEM_IO

PROC_DISK_SYSTEM_IO_RATE

PROC_DISK_SYSTEM_READ

PROC_DISK_SYSTEM_READ_CUM

PROC_DISK_SYSTEM_WRITE

PROC_DISK_SYSTEM_WRITE_CUM

PROC_DISK_VM_IO

PROC_DISK_VM_IO_RATE

PROC_DISK_VM_READ

PROC_DISK_VM_READ_CUM

PROC_DISK_VM_WRITE

PROC_DISK_VM_WRITE_CUM

PROC_DISK_WAIT_PCT


22

PROC_DISK_WAIT_PCT_CUM

PROC_DISK_WAIT_TIME

PROC_DISK_WAIT_TIME_CUM

PROC_DISPATCH

PROC_DISPATCH_CUM

PROC_EUID

PROC_FORCED_CSWITCH

PROC_FORCED_CSWITCH_CUM

PROC_FORK

PROC_FORK_CUM

PROC_GRAPHICS_WAIT_PCT

PROC_GRAPHICS_WAIT_PCT_CUM

PROC_GRAPHICS_WAIT_TIME

PROC_GRAPHICS_WAIT_TIME_CUM

PROC_GROUP_ID

PROC_GROUP_NAME

PROC_INODE_WAIT_PCT

PROC_INODE_WAIT_PCT_CUM

PROC_INODE_WAIT_TIME

PROC_INODE_WAIT_TIME_CUM

PROC_INTERRUPTS

PROC_INTERRUPTS_CUM

PROC_INTERVAL

PROC_INTERVAL_ALIVE

PROC_INTERVAL_CUM

PROC_IO_BYTE

PROC_IO_BYTE_CUM

PROC_IO_BYTE_RATE

PROC_IO_BYTE_RATE_CUM

PROC_IPC_SUBSYSTEM_WAIT_PCT

PROC_IPC_SUBSYSTEM_WAIT_PCT_CUM

PROC_IPC_SUBSYSTEM_WAIT_TIME

PROC_IPC_SUBSYSTEM_WAIT_TIME_CUM

PROC_IPC_WAIT_PCT

PROC_IPC_WAIT_PCT_CUM

PROC_IPC_WAIT_TIME


23

PROC_IPC_WAIT_TIME_CUM

PROC_JOBCTL_WAIT_PCT

PROC_JOBCTL_WAIT_PCT_CUM

PROC_JOBCTL_WAIT_TIME

PROC_JOBCTL_WAIT_TIME_CUM

PROC_LAN_WAIT_PCT

PROC_LAN_WAIT_PCT_CUM

PROC_LAN_WAIT_TIME

PROC_LAN_WAIT_TIME_CUM

PROC_MAJOR_FAULT

PROC_MAJOR_FAULT_CUM

PROC_MEM_PRIVATE_RES

PROC_MEM_RES

PROC_MEM_RES_HIGH

PROC_MEM_SHARED_RES

PROC_MEM_VFAULT_COUNT

PROC_MEM_VFAULT_COUNT_CUM

PROC_MEM_VIRT

PROC_MEM_WAIT_PCT

PROC_MEM_WAIT_PCT_CUM

PROC_MEM_WAIT_TIME

PROC_MEM_WAIT_TIME_CUM

PROC_MINOR_FAULT

PROC_MINOR_FAULT_CUM

PROC_MSG_RECEIVED

PROC_MSG_RECEIVED_CUM

PROC_MSG_SENT

PROC_MSG_SENT_CUM

PROC_MSG_WAIT_PCT

PROC_MSG_WAIT_PCT_CUM

PROC_MSG_WAIT_TIME

PROC_MSG_WAIT_TIME_CUM

PROC_NFS_WAIT_PCT

PROC_NFS_WAIT_PCT_CUM

PROC_NFS_WAIT_TIME

PROC_NFS_WAIT_TIME_CUM


24

PROC_NICE_PRI

PROC_NONDISK_LOGL_READ

PROC_NONDISK_LOGL_READ_CUM

PROC_NONDISK_LOGL_WRITE

PROC_NONDISK_LOGL_WRITE_CUM

PROC_NONDISK_PHYS_READ

PROC_NONDISK_PHYS_READ_CUM

PROC_NONDISK_PHYS_WRITE

PROC_NONDISK_PHYS_WRITE_CUM

PROC_OPEN

PROC_OPEN_CUM

PROC_OTHER_IO_WAIT_PCT

PROC_OTHER_IO_WAIT_PCT_CUM

PROC_OTHER_IO_WAIT_TIME

PROC_OTHER_IO_WAIT_TIME_CUM

PROC_OTHER_WAIT_PCT

PROC_OTHER_WAIT_PCT_CUM

PROC_OTHER_WAIT_TIME

PROC_OTHER_WAIT_TIME_CUM

PROC_PARENT_PROC_ID

PROC_PIPE_WAIT_PCT

PROC_PIPE_WAIT_PCT_CUM

PROC_PIPE_WAIT_TIME

PROC_PIPE_WAIT_TIME_CUM

PROC_PRI

PROC_PRI_WAIT_PCT

PROC_PRI_WAIT_PCT_CUM

PROC_PRI_WAIT_TIME

PROC_PRI_WAIT_TIME_CUM

PROC_PRMID

PROC_PROC_ID

PROC_PROC_NAME

PROC_RPC_WAIT_PCT

PROC_RPC_WAIT_PCT_CUM

PROC_RPC_WAIT_TIME

PROC_RPC_WAIT_TIME_CUM


25

PROC_RUN_TIME

PROC_SCHEDULER

PROC_SEM_WAIT_PCT

PROC_SEM_WAIT_PCT_CUM

PROC_SEM_WAIT_TIME

PROC_SEM_WAIT_TIME_CUM

PROC_SIGNAL

PROC_SIGNAL_CUM

PROC_SLEEP_WAIT_PCT

PROC_SLEEP_WAIT_PCT_CUM

PROC_SLEEP_WAIT_TIME

PROC_SLEEP_WAIT_TIME_CUM

PROC_SOCKET_WAIT_PCT

PROC_SOCKET_WAIT_PCT_CUM

PROC_SOCKET_WAIT_TIME

PROC_SOCKET_WAIT_TIME_CUM

PROC_STARTTIME

PROC_STATE

PROC_STOP_REASON

PROC_STOP_REASON_FLAG

PROC_STREAM_WAIT_PCT

PROC_STREAM_WAIT_PCT_CUM

PROC_STREAM_WAIT_TIME

PROC_STREAM_WAIT_TIME_CUM

PROC_SWAP

PROC_SWAP_CUM

PROC_SYS_WAIT_PCT

PROC_SYS_WAIT_PCT_CUM

PROC_SYS_WAIT_TIME

PROC_SYS_WAIT_TIME_CUM

PROC_TERM_IO_WAIT_PCT

PROC_TERM_IO_WAIT_PCT_CUM

PROC_TERM_IO_WAIT_TIME

PROC_TERM_IO_WAIT_TIME_CUM

PROC_THREAD_COUNT

PROC_THREAD_ID


26

PROC_TIME

PROC_TOP_CPU_INDEX

PROC_TOP_DISK_INDEX

PROC_TOTAL_WAIT_TIME

PROC_TOTAL_WAIT_TIME_CUM

PROC_TTY

PROC_TTY_DEV

PROC_UID

PROC_USER_NAME

PROC_USER_THREAD_ID

PROC_USRPRI

PROC_VOLUNTARY_CSWITCH

PROC_VOLUNTARY_CSWITCH_CUM

Application MetricsAPP_ACTIVE_APP

APP_ACTIVE_APP_PRM

APP_ACTIVE_PROC

APP_ALIVE_PROC

APP_COMPLETED_PROC

APP_CPU_NICE_TIME

APP_CPU_NICE_UTIL

APP_CPU_NNICE_TIME

APP_CPU_NNICE_UTIL

APP_CPU_NORMAL_TIME

APP_CPU_NORMAL_UTIL

APP_CPU_REALTIME_TIME

APP_CPU_REALTIME_UTIL

APP_CPU_SYS_MODE_TIME

APP_CPU_SYS_MODE_UTIL

APP_CPU_TOTAL_TIME

APP_CPU_TOTAL_UTIL

APP_CPU_TOTAL_UTIL_CUM

APP_CPU_USER_MODE_TIME

APP_CPU_USER_MODE_UTIL

APP_DISK_FS_IO_RATE

APP_DISK_LOGL_IO_RATE


27

APP_DISK_LOGL_READ

APP_DISK_LOGL_READ_RATE

APP_DISK_LOGL_WRITE

APP_DISK_LOGL_WRITE_RATE

APP_DISK_PHYS_IO_RATE

APP_DISK_PHYS_READ

APP_DISK_PHYS_READ_RATE

APP_DISK_PHYS_WRITE

APP_DISK_PHYS_WRITE_RATE

APP_DISK_RAW_IO_RATE

APP_DISK_SUBSYSTEM_QUEUE

APP_DISK_SUBSYSTEM_WAIT_PCT

APP_DISK_SYSTEM_IO_RATE

APP_DISK_VM_IO_RATE

APP_INTERVAL

APP_INTERVAL_CUM

APP_IO_BYTE

APP_IO_BYTE_RATE

APP_IPC_SUBSYSTEM_QUEUE

APP_IPC_SUBSYSTEM_WAIT_PCT

APP_MAJOR_FAULT

APP_MAJOR_FAULT_RATE

APP_MEM_QUEUE

APP_MEM_RES

APP_MEM_UTIL

APP_MEM_VIRT

APP_MEM_WAIT_PCT

APP_MINOR_FAULT

APP_MINOR_FAULT_RATE

APP_NAME

APP_NAME_PRM_GROUPNAME

APP_NETWORK_SUBSYSTEM_QUEUE

APP_NETWORK_SUBSYSTEM_WAIT_PCT

APP_NUM

APP_OTHER_IO_QUEUE

APP_OTHER_IO_WAIT_PCT


28

APP_PRI

APP_PRI_QUEUE

APP_PRI_WAIT_PCT

APP_PRM_CPUCAP_MODE

APP_PRM_CPU_ENTITLEMENT

APP_PRM_CPU_TOTAL_UTIL_CUM

APP_PRM_DISK_STATE

APP_PRM_GROUPID

APP_PRM_INTERVAL_CUM

APP_PRM_MEM_AVAIL

APP_PRM_MEM_ENTITLEMENT

APP_PRM_MEM_STATE

APP_PRM_MEM_UPPERBOUND

APP_PRM_MEM_UTIL

APP_PRM_STATE

APP_PRM_SUSPENDED_PROC

APP_PROC_RUN_TIME

APP_SAMPLE

APP_SEM_QUEUE

APP_SEM_WAIT_PCT

APP_SLEEP_QUEUE

APP_SLEEP_WAIT_PCT

APP_TIME

Process By File MetricsPROC_FILE_COUNT

PROC_FILE_MODE

PROC_FILE_NAME

PROC_FILE_NUMBER

PROC_FILE_OFFSET

PROC_FILE_OPEN

PROC_FILE_TYPE

By Disk MetricsBYDSK_AVG_SERVICE_TIME

BYDSK_BUS

BYDSK_CONTROLLER

BYDSK_DEVNAME


29

BYDSK_DEVNO

BYDSK_DIRNAME

BYDSK_FS_IO_RATE

BYDSK_FS_READ

BYDSK_FS_READ_RATE

BYDSK_FS_WRITE

BYDSK_FS_WRITE_RATE

BYDSK_INTERVAL

BYDSK_INTERVAL_CUM

BYDSK_LOGL_BYTE_RATE

BYDSK_LOGL_BYTE_RATE_CUM

BYDSK_LOGL_IO_RATE

BYDSK_LOGL_IO_RATE_CUM

BYDSK_LOGL_READ

BYDSK_LOGL_READ_BYTE_RATE

BYDSK_LOGL_READ_BYTE_RATE_CUM

BYDSK_LOGL_READ_RATE

BYDSK_LOGL_READ_RATE_CUM

BYDSK_LOGL_WRITE

BYDSK_LOGL_WRITE_BYTE_RATE

BYDSK_LOGL_WRITE_BYTE_RATE_CUM

BYDSK_LOGL_WRITE_RATE

BYDSK_LOGL_WRITE_RATE_CUM

BYDSK_PHYS_BYTE_RATE

BYDSK_PHYS_BYTE_RATE_CUM

BYDSK_PHYS_IO

BYDSK_PHYS_IO_RATE

BYDSK_PHYS_IO_RATE_CUM

BYDSK_PHYS_READ

BYDSK_PHYS_READ_BYTE

BYDSK_PHYS_READ_BYTE_RATE

BYDSK_PHYS_READ_BYTE_RATE_CUM

BYDSK_PHYS_READ_RATE

BYDSK_PHYS_READ_RATE_CUM

BYDSK_PHYS_WRITE

BYDSK_PHYS_WRITE_BYTE


30

BYDSK_PHYS_WRITE_BYTE_RATE

BYDSK_PHYS_WRITE_BYTE_RATE_CUM

BYDSK_PHYS_WRITE_RATE

BYDSK_PHYS_WRITE_RATE_CUM

BYDSK_PRODUCT_ID

BYDSK_QUEUE_0_UTIL

BYDSK_QUEUE_2_UTIL

BYDSK_QUEUE_4_UTIL

BYDSK_QUEUE_8_UTIL

BYDSK_QUEUE_X_UTIL

BYDSK_RAW_IO_RATE

BYDSK_RAW_READ

BYDSK_RAW_READ_RATE

BYDSK_RAW_WRITE

BYDSK_RAW_WRITE_RATE

BYDSK_REQUEST_QUEUE

BYDSK_SYSTEM_IO

BYDSK_SYSTEM_IO_RATE

BYDSK_SYSTEM_READ_RATE

BYDSK_SYSTEM_WRITE_RATE

BYDSK_TIME

BYDSK_UTIL

BYDSK_UTIL_CUM

BYDSK_VENDOR_ID

BYDSK_VM_IO

BYDSK_VM_IO_RATE

BYDSK_VM_READ_RATE

BYDSK_VM_WRITE_RATE

File System MetricsFS_BLOCK_SIZE

FS_DEVNAME

FS_DEVNO

FS_DIRNAME

FS_FILE_IO_RATE

FS_FILE_IO_RATE_CUM

FS_FRAG_SIZE


31

FS_INTERVAL

FS_INTERVAL_CUM

FS_IS_LVM

FS_LOGL_IO_RATE

FS_LOGL_IO_RATE_CUM

FS_LOGL_READ_BYTE_RATE

FS_LOGL_READ_BYTE_RATE_CUM

FS_LOGL_READ_RATE

FS_LOGL_READ_RATE_CUM

FS_LOGL_WRITE_BYTE_RATE

FS_LOGL_WRITE_BYTE_RATE_CUM

FS_LOGL_WRITE_RATE

FS_LOGL_WRITE_RATE_CUM

FS_MAX_SIZE

FS_PHYS_IO_RATE

FS_PHYS_IO_RATE_CUM

FS_PHYS_READ_BYTE_RATE

FS_PHYS_READ_BYTE_RATE_CUM

FS_PHYS_READ_RATE

FS_PHYS_READ_RATE_CUM

FS_PHYS_WRITE_BYTE_RATE

FS_PHYS_WRITE_BYTE_RATE_CUM

FS_PHYS_WRITE_RATE

FS_PHYS_WRITE_RATE_CUM

FS_SPACE_UTIL

FS_TYPE

FS_VM_IO_RATE

FS_VM_IO_RATE_CUM

Logical Volume MetricsLV_AVG_READ_SERVICE_TIME

LV_AVG_WRITE_SERVICE_TIME

LV_CACHE_HIT

LV_CACHE_MISS

LV_CACHE_QUEUE

LV_CACHE_SIZE

LV_DEVNO


32

LV_DIRNAME

LV_GROUP_NAME

LV_INTERVAL

LV_INTERVAL_CUM

LV_OPEN_LV

LV_READ_BYTE_RATE

LV_READ_BYTE_RATE_CUM

LV_READ_RATE

LV_READ_RATE_CUM

LV_TYPE

LV_WRITE_BYTE_RATE

LV_WRITE_BYTE_RATE_CUM

LV_WRITE_RATE

LV_WRITE_RATE_CUM

By Network Interface MetricsBYNETIF_COLLISION

BYNETIF_COLLISION_1_MIN_RATE

BYNETIF_COLLISION_RATE

BYNETIF_COLLISION_RATE_CUM

BYNETIF_ERROR

BYNETIF_ERROR_1_MIN_RATE

BYNETIF_ERROR_RATE

BYNETIF_ERROR_RATE_CUM

BYNETIF_INTERVAL

BYNETIF_INTERVAL_CUM

BYNETIF_IN_BYTE_RATE

BYNETIF_IN_BYTE_RATE_CUM

BYNETIF_IN_PACKET

BYNETIF_IN_PACKET_RATE

BYNETIF_IN_PACKET_RATE_CUM

BYNETIF_NAME

BYNETIF_NET_MTU

BYNETIF_NET_SPEED

BYNETIF_NET_TYPE

BYNETIF_OUT_BYTE_RATE

BYNETIF_OUT_BYTE_RATE_CUM


33

BYNETIF_OUT_PACKET

BYNETIF_OUT_PACKET_RATE

BYNETIF_OUT_PACKET_RATE_CUM

BYNETIF_PACKET_RATE

BYNETIF_QUEUE

By Swap MetricsBYSWP_SWAP_PRI

BYSWP_SWAP_SPACE_AVAIL

BYSWP_SWAP_SPACE_NAME

BYSWP_SWAP_SPACE_USED

BYSWP_SWAP_TYPE

By CPU MetricsBYCPU_CPU_CSWITCH_TIME

BYCPU_CPU_CSWITCH_TIME_CUM

BYCPU_CPU_CSWITCH_UTIL

BYCPU_CPU_CSWITCH_UTIL_CUM

BYCPU_CPU_INTERRUPT_TIME

BYCPU_CPU_INTERRUPT_TIME_CUM

BYCPU_CPU_INTERRUPT_UTIL

BYCPU_CPU_INTERRUPT_UTIL_CUM

BYCPU_CPU_NICE_TIME

BYCPU_CPU_NICE_TIME_CUM

BYCPU_CPU_NICE_UTIL

BYCPU_CPU_NICE_UTIL_CUM

BYCPU_CPU_NNICE_TIME

BYCPU_CPU_NNICE_TIME_CUM

BYCPU_CPU_NNICE_UTIL

BYCPU_CPU_NNICE_UTIL_CUM

BYCPU_CPU_NORMAL_TIME

BYCPU_CPU_NORMAL_TIME_CUM

BYCPU_CPU_NORMAL_UTIL

BYCPU_CPU_NORMAL_UTIL_CUM

BYCPU_CPU_REALTIME_TIME

BYCPU_CPU_REALTIME_TIME_CUM

BYCPU_CPU_REALTIME_UTIL

BYCPU_CPU_REALTIME_UTIL_CUM


34

BYCPU_CPU_SYSCALL_TIME

BYCPU_CPU_SYSCALL_TIME_CUM

BYCPU_CPU_SYSCALL_UTIL

BYCPU_CPU_SYSCALL_UTIL_CUM

BYCPU_CPU_SYS_MODE_TIME

BYCPU_CPU_SYS_MODE_TIME_CUM

BYCPU_CPU_SYS_MODE_UTIL

BYCPU_CPU_SYS_MODE_UTIL_CUM

BYCPU_CPU_TOTAL_UTIL

BYCPU_CPU_TOTAL_UTIL_CUM

BYCPU_CPU_TRAP_TIME

BYCPU_CPU_TRAP_TIME_CUM

BYCPU_CPU_TRAP_UTIL

BYCPU_CPU_TRAP_UTIL_CUM

BYCPU_CPU_USER_MODE_TIME

BYCPU_CPU_USER_MODE_TIME_CUM

BYCPU_CPU_USER_MODE_UTIL

BYCPU_CPU_USER_MODE_UTIL_CUM

BYCPU_CPU_VFAULT_TIME

BYCPU_CPU_VFAULT_TIME_CUM

BYCPU_CPU_VFAULT_UTIL

BYCPU_CPU_VFAULT_UTIL_CUM

BYCPU_CSWITCH

BYCPU_CSWITCH_CUM

BYCPU_CSWITCH_RATE

BYCPU_CSWITCH_RATE_CUM

BYCPU_ID

BYCPU_INTERRUPT_RATE

BYCPU_INTERRUPT_STATE

BYCPU_LAST_PROC_ID

BYCPU_LAST_THREAD_ID

BYCPU_LAST_USER_THREAD_ID

BYCPU_RUN_QUEUE_15_MIN



BYCPU_STATE


35

Process By Memory Region MetricsPROC_REGION_FILENAME

PROC_REGION_LOCKED

PROC_REGION_PAGE_COUNT_1_4KB




PROC_REGION_PAGE_COUNT_5_1MB




PROC_REGION_PAGE_SIZE_HINT

PROC_REGION_PRIVATE_SHARED_FLAG

PROC_REGION_REF_COUNT

PROC_REGION_RES

PROC_REGION_RES_DATA

PROC_REGION_RES_OTHER

PROC_REGION_RES_SHMEM

PROC_REGION_RES_STACK

PROC_REGION_RES_TEXT

PROC_REGION_TYPE

PROC_REGION_VIRT

PROC_REGION_VIRT_ADDRS

PROC_REGION_VIRT_DATA

PROC_REGION_VIRT_OTHER

PROC_REGION_VIRT_SHMEM

PROC_REGION_VIRT_STACK

PROC_REGION_VIRT_TEXT

By NFS MetricsBYNFS_CLIENT_PHYS_TIME

BYNFS_CLIENT_PHYS_TIME_CUM

BYNFS_CLIENT_READ_BYTE_RATE

BYNFS_CLIENT_READ_BYTE_RATE_CUM

BYNFS_CLIENT_READ_RATE

BYNFS_CLIENT_READ_RATE_CUM

BYNFS_CLIENT_SERVICE


36

BYNFS_CLIENT_SERVICE_CUM

BYNFS_CLIENT_SERVICE_QUEUE

BYNFS_CLIENT_SERVICE_QUEUE_CUM

BYNFS_CLIENT_SERVICE_TIME

BYNFS_CLIENT_SERVICE_TIME_CUM

BYNFS_CLIENT_WRITE_BYTE_RATE

BYNFS_CLIENT_WRITE_BYTE_RATE_CUM

BYNFS_CLIENT_WRITE_RATE

BYNFS_CLIENT_WRITE_RATE_CUM

BYNFS_HOSTNAME

BYNFS_HOST_IP_ADDRESS

BYNFS_INTERVAL

BYNFS_INTERVAL_CUM

BYNFS_LAST_PROC_ID

BYNFS_SERVER_READ_BYTE_RATE

BYNFS_SERVER_READ_BYTE_RATE_CUM

BYNFS_SERVER_READ_RATE

BYNFS_SERVER_READ_RATE_CUM

BYNFS_SERVER_SERVICE

BYNFS_SERVER_SERVICE_CUM

BYNFS_SERVER_SERVICE_TIME

BYNFS_SERVER_SERVICE_TIME_CUM

BYNFS_SERVER_WRITE_BYTE_RATE

BYNFS_SERVER_WRITE_BYTE_RATE_CUM

BYNFS_SERVER_WRITE_RATE

BYNFS_SERVER_WRITE_RATE_CUM

By NFS Operation MetricsBYNFSOP_CLIENT_COUNT

BYNFSOP_CLIENT_COUNT_CUM

BYNFSOP_CLIENT_TIME

BYNFSOP_CLIENT_TIME_CUM

BYNFSOP_INTERVAL

BYNFSOP_INTERVAL_CUM

BYNFSOP_NAME

BYNFSOP_SERVER_COUNT

BYNFSOP_SERVER_COUNT_CUM


37

BYNFSOP_SERVER_TIME

BYNFSOP_SERVER_TIME_CUM

By Operation MetricsBYOP_CLIENT_COUNT

BYOP_CLIENT_COUNT_CUM

BYOP_INTERVAL

BYOP_INTERVAL_CUM

BYOP_NAME

BYOP_SERVER_COUNT

BYOP_SERVER_COUNT_CUM

System Call MetricsSYSCALL_ACTIVE_CUM

SYSCALL_CALL_COUNT

SYSCALL_CALL_COUNT_CUM

SYSCALL_CALL_ID

SYSCALL_CALL_NAME

SYSCALL_CALL_RATE

SYSCALL_CALL_RATE_CUM

SYSCALL_CPU_TOTAL_TIME

SYSCALL_CPU_TOTAL_TIME_CUM

SYSCALL_INTERVAL

SYSCALL_INTERVAL_CUM

By Disk Detail MetricsBYDSKDETAIL_LABEL

BYDSKDETAIL_NAME

File System Detail MetricsFSDETAIL_LABEL

FSDETAIL_NAME

Logical Volume Detail MetricsLVDETAIL_LABEL

LVDETAIL_NAME

Transaction MetricsTT_ABORT

TT_ABORT_CUM

TT_ABORT_WALL_TIME


38

TT_ABORT_WALL_TIME_CUM

TT_APPNO

TT_APP_NAME

TT_CACHE_WAIT_TIME_PER_TRAN

TT_CACHE_WAIT_TIME_PER_TRAN_CUM

TT_CDFS_WAIT_TIME_PER_TRAN

TT_CDFS_WAIT_TIME_PER_TRAN_CUM

TT_CLIENT_CORRELATOR_COUNT

TT_COUNT

TT_COUNT_CUM

TT_CPU_CSWITCH_TIME_PER_TRAN

TT_CPU_CSWITCH_TIME_PER_TRAN_CUM

TT_CPU_INTERRUPT_TIME_PER_TRAN

TT_CPU_INTERRUPT_TIME_PER_TRAN_CUM

TT_CPU_NICE_TIME_PER_TRAN

TT_CPU_NICE_TIME_PER_TRAN_CUM

TT_CPU_NNICE_TIME_PER_TRAN

TT_CPU_NNICE_TIME_PER_TRAN_CUM

TT_CPU_NORMAL_TIME_PER_TRAN

TT_CPU_NORMAL_TIME_PER_TRAN_CUM

TT_CPU_REALTIME_TIME_PER_TRAN

TT_CPU_REALTIME_TIME_PER_TRAN_CUM

TT_CPU_SYSCALL_TIME_PER_TRAN

TT_CPU_SYSCALL_TIME_PER_TRAN_CUM

TT_CPU_SYS_MODE_TIME_PER_TRAN

TT_CPU_SYS_MODE_TIME_PER_TRAN_CUM

TT_CPU_TOTAL_TIME_PER_TRAN

TT_CPU_TOTAL_TIME_PER_TRAN_CUM

TT_CPU_USER_MODE_TIME_PER_TRAN

TT_CPU_USER_MODE_TIME_PER_TRAN_CUM

TT_DISK_FS_READ_PER_TRAN

TT_DISK_FS_READ_PER_TRAN_CUM

TT_DISK_FS_WRITE_PER_TRAN

TT_DISK_FS_WRITE_PER_TRAN_CUM

TT_DISK_LOGL_IO_PER_TRAN

TT_DISK_LOGL_IO_PER_TRAN_CUM


39

TT_DISK_LOGL_READ_PER_TRAN

TT_DISK_LOGL_READ_PER_TRAN_CUM

TT_DISK_LOGL_WRITE_PER_TRAN

TT_DISK_LOGL_WRITE_PER_TRAN_CUM

TT_DISK_PHYS_IO_PER_TRAN

TT_DISK_PHYS_IO_PER_TRAN_CUM

TT_DISK_PHYS_READ_PER_TRAN

TT_DISK_PHYS_READ_PER_TRAN_CUM

TT_DISK_PHYS_WRITE_PER_TRAN

TT_DISK_PHYS_WRITE_PER_TRAN_CUM

TT_DISK_RAW_READ_PER_TRAN

TT_DISK_RAW_READ_PER_TRAN_CUM

TT_DISK_RAW_WRITE_PER_TRAN

TT_DISK_RAW_WRITE_PER_TRAN_CUM

TT_DISK_SYSTEM_READ_PER_TRAN

TT_DISK_SYSTEM_READ_PER_TRAN_CUM

TT_DISK_SYSTEM_WRITE_PER_TRAN

TT_DISK_SYSTEM_WRITE_PER_TRAN_CUM

TT_DISK_VM_READ_PER_TRAN

TT_DISK_VM_READ_PER_TRAN_CUM

TT_DISK_VM_WRITE_PER_TRAN

TT_DISK_VM_WRITE_PER_TRAN_CUM

TT_DISK_WAIT_TIME_PER_TRAN

TT_DISK_WAIT_TIME_PER_TRAN_CUM

TT_FAILED

TT_FAILED_CUM

TT_FAILED_WALL_TIME

TT_FAILED_WALL_TIME_CUM

TT_GOLDENRESOURCE_INTERVAL

TT_GOLDENRESOURCE_INTERVAL_CUM

TT_GRAPHICS_WAIT_TIME_PER_TRAN

TT_GRAPHICS_WAIT_TIME_PER_TRAN_CUM

TT_INFO

TT_INODE_WAIT_TIME_PER_TRAN

TT_INODE_WAIT_TIME_PER_TRAN_CUM

TT_INPROGRESS_COUNT


40

TT_INTERVAL

TT_INTERVAL_CUM

TT_IPC_WAIT_TIME_PER_TRAN

TT_IPC_WAIT_TIME_PER_TRAN_CUM

TT_JOBCTL_WAIT_TIME_PER_TRAN

TT_JOBCTL_WAIT_TIME_PER_TRAN_CUM

TT_LAN_WAIT_TIME_PER_TRAN

TT_LAN_WAIT_TIME_PER_TRAN_CUM

TT_MEASUREMENT_COUNT

TT_MEM_WAIT_TIME_PER_TRAN

TT_MEM_WAIT_TIME_PER_TRAN_CUM

TT_MSG_WAIT_TIME_PER_TRAN

TT_MSG_WAIT_TIME_PER_TRAN_CUM

TT_NAME

TT_NFS_WAIT_TIME_PER_TRAN

TT_NFS_WAIT_TIME_PER_TRAN_CUM

TT_OTHER_IO_WAIT_TIME_PER_TRAN

TT_OTHER_IO_WAIT_TIME_PER_TRAN_CUM

TT_OTHER_WAIT_TIME_PER_TRAN

TT_OTHER_WAIT_TIME_PER_TRAN_CUM

TT_PIPE_WAIT_TIME_PER_TRAN

TT_PIPE_WAIT_TIME_PER_TRAN_CUM

TT_PRI_WAIT_TIME_PER_TRAN

TT_PRI_WAIT_TIME_PER_TRAN_CUM

TT_RESOURCE_INTERVAL

TT_RESOURCE_INTERVAL_CUM

TT_RPC_WAIT_TIME_PER_TRAN

TT_RPC_WAIT_TIME_PER_TRAN_CUM

TT_SEM_WAIT_TIME_PER_TRAN

TT_SEM_WAIT_TIME_PER_TRAN_CUM

TT_SLEEP_WAIT_TIME_PER_TRAN

TT_SLEEP_WAIT_TIME_PER_TRAN_CUM

TT_SLO_COUNT

TT_SLO_COUNT_CUM

TT_SLO_THRESHOLD

TT_SOCKET_WAIT_TIME_PER_TRAN


41

TT_SOCKET_WAIT_TIME_PER_TRAN_CUM

TT_STREAM_WAIT_TIME_PER_TRAN

TT_STREAM_WAIT_TIME_PER_TRAN_CUM

TT_SYS_WAIT_TIME_PER_TRAN

TT_SYS_WAIT_TIME_PER_TRAN_CUM

TT_TERM_IO_WAIT_TIME_PER_TRAN

TT_TERM_IO_WAIT_TIME_PER_TRAN_CUM

TT_TOTAL_WAIT_TIME_PER_TRAN

TT_TOTAL_WAIT_TIME_PER_TRAN_CUM

TT_TRAN_ID

TT_UNAME

TT_UPDATE

TT_UPDATE_CUM

TT_WALL_TIME

TT_WALL_TIME_CUM

TT_WALL_TIME_PER_TRAN

TT_WALL_TIME_PER_TRAN_CUM

Transaction Measurement Section MetricsTTBIN_TRANS_COUNT

TTBIN_TRANS_COUNT_CUM

TTBIN_UPPER_RANGE

By Process System Call MetricsPROCSYSCALL_ACTIVE_CUM

PROCSYSCALL_CALL_COUNT

PROCSYSCALL_CALL_COUNT_CUM

PROCSYSCALL_CALL_ID

PROCSYSCALL_CALL_NAME

PROCSYSCALL_CALL_RATE

PROCSYSCALL_CALL_RATE_CUM

PROCSYSCALL_INTERVAL

PROCSYSCALL_INTERVAL_CUM

PROCSYSCALL_TOTAL_TIME

PROCSYSCALL_TOTAL_TIME_CUM

Thread MetricsTHREAD_ACTIVE_THREAD

THREAD_APP_ID


42

THREAD_APP_NAME

THREAD_CACHE_WAIT_PCT

THREAD_CACHE_WAIT_PCT_CUM

THREAD_CACHE_WAIT_TIME

THREAD_CACHE_WAIT_TIME_CUM

THREAD_CDFS_WAIT_PCT

THREAD_CDFS_WAIT_PCT_CUM

THREAD_CDFS_WAIT_TIME

THREAD_CDFS_WAIT_TIME_CUM

THREAD_CLOSE

THREAD_CLOSE_CUM

THREAD_CPU_CSWITCH_TIME

THREAD_CPU_CSWITCH_TIME_CUM

THREAD_CPU_CSWITCH_UTIL

THREAD_CPU_CSWITCH_UTIL_CUM

THREAD_CPU_INTERRUPT_TIME

THREAD_CPU_INTERRUPT_TIME_CUM

THREAD_CPU_INTERRUPT_UTIL

THREAD_CPU_INTERRUPT_UTIL_CUM

THREAD_CPU_LAST_USED

THREAD_CPU_NICE_TIME

THREAD_CPU_NICE_TIME_CUM

THREAD_CPU_NICE_UTIL

THREAD_CPU_NICE_UTIL_CUM

THREAD_CPU_NNICE_TIME

THREAD_CPU_NNICE_TIME_CUM

THREAD_CPU_NNICE_UTIL

THREAD_CPU_NNICE_UTIL_CUM

THREAD_CPU_NORMAL_TIME

THREAD_CPU_NORMAL_TIME_CUM

THREAD_CPU_NORMAL_UTIL

THREAD_CPU_NORMAL_UTIL_CUM

THREAD_CPU_REALTIME_TIME

THREAD_CPU_REALTIME_TIME_CUM

THREAD_CPU_REALTIME_UTIL

THREAD_CPU_REALTIME_UTIL_CUM


43

THREAD_CPU_SWITCHES

THREAD_CPU_SWITCHES_CUM

THREAD_CPU_SYSCALL_TIME

THREAD_CPU_SYSCALL_TIME_CUM

THREAD_CPU_SYSCALL_UTIL

THREAD_CPU_SYSCALL_UTIL_CUM

THREAD_CPU_SYS_MODE_TIME

THREAD_CPU_SYS_MODE_TIME_CUM

THREAD_CPU_SYS_MODE_UTIL

THREAD_CPU_SYS_MODE_UTIL_CUM

THREAD_CPU_TOTAL_TIME

THREAD_CPU_TOTAL_TIME_CUM

THREAD_CPU_TOTAL_UTIL

THREAD_CPU_TOTAL_UTIL_CUM

THREAD_CPU_TRAP_COUNT

THREAD_CPU_TRAP_COUNT_CUM

THREAD_CPU_USER_MODE_TIME

THREAD_CPU_USER_MODE_TIME_CUM

THREAD_CPU_USER_MODE_UTIL

THREAD_CPU_USER_MODE_UTIL_CUM

THREAD_DISK_FS_READ

THREAD_DISK_FS_READ_CUM

THREAD_DISK_FS_READ_RATE

THREAD_DISK_FS_WRITE

THREAD_DISK_FS_WRITE_CUM

THREAD_DISK_FS_WRITE_RATE

THREAD_DISK_LOGL_IO

THREAD_DISK_LOGL_IO_CUM

THREAD_DISK_LOGL_IO_RATE

THREAD_DISK_LOGL_IO_RATE_CUM

THREAD_DISK_LOGL_READ

THREAD_DISK_LOGL_READ_CUM

THREAD_DISK_LOGL_READ_RATE

THREAD_DISK_LOGL_WRITE

THREAD_DISK_LOGL_WRITE_CUM

THREAD_DISK_LOGL_WRITE_RATE


44

THREAD_DISK_PHYS_IO_RATE

THREAD_DISK_PHYS_IO_RATE_CUM

THREAD_DISK_PHYS_READ

THREAD_DISK_PHYS_READ_CUM

THREAD_DISK_PHYS_READ_RATE

THREAD_DISK_PHYS_WRITE

THREAD_DISK_PHYS_WRITE_CUM

THREAD_DISK_PHYS_WRITE_RATE

THREAD_DISK_RAW_READ

THREAD_DISK_RAW_READ_CUM

THREAD_DISK_RAW_READ_RATE

THREAD_DISK_RAW_WRITE

THREAD_DISK_RAW_WRITE_CUM

THREAD_DISK_RAW_WRITE_RATE

THREAD_DISK_REM_LOGL_READ

THREAD_DISK_REM_LOGL_READ_CUM

THREAD_DISK_REM_LOGL_READ_RATE

THREAD_DISK_REM_LOGL_WRITE

THREAD_DISK_REM_LOGL_WRITE_CUM

THREAD_DISK_REM_LOGL_WRITE_RATE

THREAD_DISK_REM_PHYS_READ

THREAD_DISK_REM_PHYS_READ_CUM

THREAD_DISK_REM_PHYS_READ_RATE

THREAD_DISK_REM_PHYS_WRITE

THREAD_DISK_REM_PHYS_WRITE_CUM

THREAD_DISK_REM_PHYS_WRITE_RATE

THREAD_DISK_SUBSYSTEM_WAIT_PCT

THREAD_DISK_SUBSYSTEM_WAIT_PCT_CUM

THREAD_DISK_SUBSYSTEM_WAIT_TIME

THREAD_DISK_SUBSYSTEM_WAIT_TIME_CUM

THREAD_DISK_SYSTEM_IO

THREAD_DISK_SYSTEM_IO_RATE

THREAD_DISK_SYSTEM_READ

THREAD_DISK_SYSTEM_READ_CUM

THREAD_DISK_SYSTEM_WRITE

THREAD_DISK_SYSTEM_WRITE_CUM


45

THREAD_DISK_VM_IO

THREAD_DISK_VM_IO_RATE

THREAD_DISK_VM_READ

THREAD_DISK_VM_READ_CUM

THREAD_DISK_VM_WRITE

THREAD_DISK_VM_WRITE_CUM

THREAD_DISK_WAIT_PCT

THREAD_DISK_WAIT_PCT_CUM

THREAD_DISK_WAIT_TIME

THREAD_DISK_WAIT_TIME_CUM

THREAD_DISPATCH

THREAD_DISPATCH_CUM

THREAD_EUID

THREAD_FORCED_CSWITCH

THREAD_FORCED_CSWITCH_CUM

THREAD_FORK

THREAD_FORK_CUM

THREAD_GRAPHICS_WAIT_PCT

THREAD_GRAPHICS_WAIT_PCT_CUM

THREAD_GRAPHICS_WAIT_TIME

THREAD_GRAPHICS_WAIT_TIME_CUM

THREAD_GROUP_ID

THREAD_GROUP_NAME

THREAD_INODE_WAIT_PCT

THREAD_INODE_WAIT_PCT_CUM

THREAD_INODE_WAIT_TIME

THREAD_INODE_WAIT_TIME_CUM

THREAD_INTERRUPTS

THREAD_INTERRUPTS_CUM

THREAD_INTERVAL

THREAD_INTERVAL_ALIVE

THREAD_INTERVAL_CUM

THREAD_IO_BYTE

THREAD_IO_BYTE_CUM

THREAD_IO_BYTE_RATE

THREAD_IO_BYTE_RATE_CUM


46

THREAD_IPC_SUBSYSTEM_WAIT_PCT

THREAD_IPC_SUBSYSTEM_WAIT_PCT_CUM

THREAD_IPC_SUBSYSTEM_WAIT_TIME

THREAD_IPC_SUBSYSTEM_WAIT_TIME_CUM

THREAD_IPC_WAIT_PCT

THREAD_IPC_WAIT_PCT_CUM

THREAD_IPC_WAIT_TIME

THREAD_IPC_WAIT_TIME_CUM

THREAD_JOBCTL_WAIT_PCT

THREAD_JOBCTL_WAIT_PCT_CUM

THREAD_JOBCTL_WAIT_TIME

THREAD_JOBCTL_WAIT_TIME_CUM

THREAD_LAN_WAIT_PCT

THREAD_LAN_WAIT_PCT_CUM

THREAD_LAN_WAIT_TIME

THREAD_LAN_WAIT_TIME_CUM

THREAD_MAJOR_FAULT

THREAD_MAJOR_FAULT_CUM

THREAD_MEM_PRIVATE_RES

THREAD_MEM_RES

THREAD_MEM_RES_HIGH

THREAD_MEM_SHARED_RES

THREAD_MEM_VFAULT_COUNT

THREAD_MEM_VFAULT_COUNT_CUM

THREAD_MEM_VIRT

THREAD_MEM_WAIT_PCT

THREAD_MEM_WAIT_PCT_CUM

THREAD_MEM_WAIT_TIME

THREAD_MEM_WAIT_TIME_CUM

THREAD_MINOR_FAULT

THREAD_MINOR_FAULT_CUM

THREAD_MSG_RECEIVED

THREAD_MSG_RECEIVED_CUM

THREAD_MSG_SENT

THREAD_MSG_SENT_CUM

THREAD_MSG_WAIT_PCT


47

THREAD_MSG_WAIT_PCT_CUM

THREAD_MSG_WAIT_TIME

THREAD_MSG_WAIT_TIME_CUM

THREAD_NFS_WAIT_PCT

THREAD_NFS_WAIT_PCT_CUM

THREAD_NFS_WAIT_TIME

THREAD_NFS_WAIT_TIME_CUM

THREAD_NICE_PRI

THREAD_NONDISK_LOGL_READ

THREAD_NONDISK_LOGL_READ_CUM

THREAD_NONDISK_LOGL_WRITE

THREAD_NONDISK_LOGL_WRITE_CUM

THREAD_NONDISK_PHYS_READ

THREAD_NONDISK_PHYS_READ_CUM

THREAD_NONDISK_PHYS_WRITE

THREAD_NONDISK_PHYS_WRITE_CUM

THREAD_OPEN

THREAD_OPEN_CUM

THREAD_OTHER_IO_WAIT_PCT

THREAD_OTHER_IO_WAIT_PCT_CUM

THREAD_OTHER_IO_WAIT_TIME

THREAD_OTHER_IO_WAIT_TIME_CUM

THREAD_OTHER_WAIT_PCT

THREAD_OTHER_WAIT_PCT_CUM

THREAD_OTHER_WAIT_TIME

THREAD_OTHER_WAIT_TIME_CUM

THREAD_PARENT_PROC_ID

THREAD_PIPE_WAIT_PCT

THREAD_PIPE_WAIT_PCT_CUM

THREAD_PIPE_WAIT_TIME

THREAD_PIPE_WAIT_TIME_CUM

THREAD_PRI

THREAD_PRI_WAIT_PCT

THREAD_PRI_WAIT_PCT_CUM

THREAD_PRI_WAIT_TIME

THREAD_PRI_WAIT_TIME_CUM


48

THREAD_PRMID

THREAD_PROC_ID

THREAD_PROC_NAME

THREAD_RPC_WAIT_PCT

THREAD_RPC_WAIT_PCT_CUM

THREAD_RPC_WAIT_TIME

THREAD_RPC_WAIT_TIME_CUM

THREAD_RUN_TIME

THREAD_SCHEDULER

THREAD_SEM_WAIT_PCT

THREAD_SEM_WAIT_PCT_CUM

THREAD_SEM_WAIT_TIME

THREAD_SEM_WAIT_TIME_CUM

THREAD_SIGNAL

THREAD_SIGNAL_CUM

THREAD_SLEEP_WAIT_PCT

THREAD_SLEEP_WAIT_PCT_CUM

THREAD_SLEEP_WAIT_TIME

THREAD_SLEEP_WAIT_TIME_CUM

THREAD_SOCKET_WAIT_PCT

THREAD_SOCKET_WAIT_PCT_CUM

THREAD_SOCKET_WAIT_TIME

THREAD_SOCKET_WAIT_TIME_CUM

THREAD_STARTTIME

THREAD_STATE

THREAD_STOP_REASON

THREAD_STOP_REASON_FLAG

THREAD_STREAM_WAIT_PCT

THREAD_STREAM_WAIT_PCT_CUM

THREAD_STREAM_WAIT_TIME

THREAD_STREAM_WAIT_TIME_CUM

THREAD_SWAP

THREAD_SWAP_CUM

THREAD_SYS_WAIT_PCT

THREAD_SYS_WAIT_PCT_CUM

THREAD_SYS_WAIT_TIME


49

THREAD_SYS_WAIT_TIME_CUM

THREAD_TERM_IO_WAIT_PCT

THREAD_TERM_IO_WAIT_PCT_CUM

THREAD_TERM_IO_WAIT_TIME

THREAD_TERM_IO_WAIT_TIME_CUM

THREAD_THREAD_COUNT

THREAD_THREAD_ID

THREAD_TIME

THREAD_TOP_CPU_INDEX

THREAD_TOP_DISK_INDEX

THREAD_TOTAL_WAIT_TIME

THREAD_TOTAL_WAIT_TIME_CUM

THREAD_TTY

THREAD_TTY_DEV

THREAD_UID

THREAD_USER_NAME

THREAD_USER_THREAD_ID

THREAD_USRPRI

THREAD_VOLUNTARY_CSWITCH

THREAD_VOLUNTARY_CSWITCH_CUM

Network by Logical Detail MetricsBYNETIF_LOGL_INTERVAL

BYNETIF_LOGL_INTERVAL_CUM

BYNETIF_LOGL_IN_PACKET

BYNETIF_LOGL_IN_PACKET_RATE

BYNETIF_LOGL_IN_PACKET_RATE_CUM

BYNETIF_LOGL_IP_ADDRESS

BYNETIF_LOGL_NAME

BYNETIF_LOGL_OUT_PACKET

BYNETIF_LOGL_OUT_PACKET_RATE

BYNETIF_LOGL_OUT_PACKET_RATE_CUM

Transaction Client MetricsTT_CLIENT_ABORT

TT_CLIENT_ABORT_CUM

TT_CLIENT_ABORT_WALL_TIME

TT_CLIENT_ABORT_WALL_TIME_CUM


50

TT_CLIENT_ADDRESS

TT_CLIENT_ADDRESS_FORMAT

TT_CLIENT_TRAN_ID

TT_CLIENT_COUNT

TT_CLIENT_COUNT_CUM

TT_CLIENT_FAILED

TT_CLIENT_FAILED_CUM

TT_CLIENT_FAILED_WALL_TIME

TT_CLIENT_FAILED_WALL_TIME_CUM

TT_CLIENT_INTERVAL

TT_CLIENT_INTERVAL_CUM

TT_CLIENT_SLO_COUNT

TT_CLIENT_SLO_COUNT_CUM

TT_CLIENT_UPDATE

TT_CLIENT_UPDATE_CUM

TT_CLIENT_WALL_TIME

TT_CLIENT_WALL_TIME_CUM

TT_CLIENT_WALL_TIME_PER_TRAN

TT_CLIENT_WALL_TIME_PER_TRAN_CUM

Transaction Instance MetricsTT_INSTANCE_ID

TT_INSTANCE_PROC_ID

TT_INSTANCE_START_TIME

TT_INSTANCE_STOP_TIME

TT_INSTANCE_THREAD_ID

TT_INSTANCE_UPDATE_COUNT

TT_INSTANCE_UPDATE_TIME

TT_INSTANCE_WALL_TIME

Transaction User Defined Measurement MetricsTT_USER_MEASUREMENT_AVG

TT_USER_MEASUREMENT_COUNT

TT_USER_MEASUREMENT_MAX

TT_USER_MEASUREMENT_MIN

TT_USER_MEASUREMENT_NAME

TT_USER_MEASUREMENT_STRING1024_VALUE

TT_USER_MEASUREMENT_STRING32_VALUE


51

TT_USER_MEASUREMENT_TYPE

TT_USER_MEASUREMENT_VALUE

Transaction Client User Defined MeasurementMetricsTT_CLIENT_USER_MEASUREMENT_AVG

TT_CLIENT_USER_MEASUREMENT_COUNT

TT_CLIENT_USER_MEASUREMENT_MAX

TT_CLIENT_USER_MEASUREMENT_MIN

TT_CLIENT_USER_MEASUREMENT_NAME

TT_CLIENT_USER_MEASUREMENT_STRING1024_VALUE

TT_CLIENT_USER_MEASUREMENT_STRING32_VALUE

TT_CLIENT_USER_MEASUREMENT_TYPE

TT_CLIENT_USER_MEASUREMENT_VALUE

Transaction Instance User Defined MeasurementMetricsTT_INSTANCE_USER_MEASUREMENT_AVG

TT_INSTANCE_USER_MEASUREMENT_COUNT

TT_INSTANCE_USER_MEASUREMENT_MAX

TT_INSTANCE_USER_MEASUREMENT_MIN

TT_INSTANCE_USER_MEASUREMENT_NAME

TT_INSTANCE_USER_MEASUREMENT_STRING1024_VALUE

TT_INSTANCE_USER_MEASUREMENT_STRING32_VALUE

TT_INSTANCE_USER_MEASUREMENT_TYPE

TT_INSTANCE_USER_MEASUREMENT_VALUE

PRM By Volume Group MetricsPRM_BYVG_GROUP_ENTITLEMENT

PRM_BYVG_GROUP_UTIL

PRM_BYVG_INTERVAL

PRM_BYVG_INTERVAL_CUM

PRM_BYVG_PRM_GROUPID

PRM_BYVG_PRM_GROUPNAME

PRM_BYVG_REQUEST

PRM_BYVG_REQUEST_CUM

PRM_BYVG_REQUEST_QUEUE

PRM_BYVG_TRANSFER

PRM_BYVG_TRANSFER_CUM


52

Metric Definitions

APP_ACTIVE_APPThe number of applications that had processes active (consuming cpuresources) during the interval.

APP_ACTIVE_APP_PRMThe number of PRM groups with at least one process that had activityduring the interval.

APP_ACTIVE_PROCAn active process is one that exists and consumes some CPU time.APP_ACTIVE_PROC is the sum of the alive-process-time/interval-timeratios of every process belonging to an application that is active (uses anyCPU time) during an interval.

The following diagram of a four second interval showing two processes, Aand B, for an application should be used to understand the abovedefinition. Note the difference between active processes, which consumeCPU time, and alive processes which merely exist on the system.

----------- Seconds -----------

1 2 3 4

Proc

---- ---- ---- ---- ----

A live live live live

B live/CPU live/CPU live dead

Process A is alive for the entire four second interval, but consumes noCPU. A's contribution to APP_ALIVE_PROC is 4*1/4. A contributes 0*1/4to APP_ACTIVE_PROC. B's contribution to APP_ALIVE_PROC is 3*1/4.B contributes 2*1/4 to APP_ACTIVE_PROC. Thus, for this interval,APP_ACTIVE_PROC equals 0.5 and APP_ALIVE_PROC equals 1.75.

Because a process may be alive but not active, APP_ACTIVE_PROC willalways be less than or equal to APP_ALIVE_PROC.

This metric indicates the number of processes in an application group thatare competing for the CPU. This metric is useful, along with other metrics,for comparing loads placed on the system by different groups ofprocesses.

GlancePlus Dictionary of Performance MetricsMetric Definitions

53

APP_ALIVE_PROCAn alive process is one that exists on the system. APP_ALIVE_PROC isthe sum of the alive-process-time/interval-time ratios for every processbelonging to a given application.

The following diagram of a four second interval showing two processes, Aand B, for an application should be used to understand the abovedefinition. Note the difference between active processes, which consumeCPU time, and alive processes which merely exist on the system.

----------- Seconds -----------

1 2 3 4

Proc

---- ---- ---- ---- ----



Process A is alive for the entire four second interval but consumes noCPU. A's contribution to APP_ALIVE_PROC is 4*1/4. A contributes 0*1/4to APP_ACTIVE_PROC. B's contribution to APP_ALIVE_PROC is 3*1/4.B contributes 2*1/4 to APP_ACTIVE_PROC. Thus, for this interval,APP_ACTIVE_PROC equals 0.5 and APP_ALIVE_PROC equals 1.75.

Because a process may be alive but not active, APP_ACTIVE_PROC willalways be less than or equal to APP_ALIVE_PROC.

APP_COMPLETED_PROCThe number of processes in this group that completed during the interval.

APP_CPU_NICE_TIMEThe time, in seconds, that processes in this group were using the CPU inuser mode at a nice priority during the interval. The NICE metrics includepositive nice value CPU time only. Negative nice value CPU is broken outinto NNICE (negative nice) metrics. Positive nice values range from 20 to39. Negative nice values range from 0 to 19.

On a system with multiple CPUs, this metric is normalized. That is, theCPU used over all processors is divided by the number of processorsonline. This represents the usage of the total processing capacityavailable.

APP_CPU_NICE_UTILThe percentage of time that processes in this group were using the CPU inuser mode at a nice priority during the interval. The NICE metrics includepositive nice value CPU time only. Negative nice value CPU is broken out


54

into NNICE (negative nice) metrics. Positive nice values range from 20 to39. Negative nice values range from 0 to 19.


APP_CPU_NNICE_TIMEThe time, in seconds, that processes in this group were using the CPU inuser mode at a nice priority computed from using negative nice valuesduring the interval.


APP_CPU_NNICE_UTILThe percentage of time that processes in this group were using the CPU inuser mode at a nice priority computed from using negative nice valuesduring the interval.

The NICE metrics include positive nice value CPU time only. Negativenice value CPU is broken out into NNICE (negative nice) metrics. Positivenice values range from 20 to 39. Negative nice values range from 0 to 19.


APP_CPU_NORMAL_TIMEThe time, in seconds, that processes in this group were in user mode at anormal priority during the interval. Normal priority user mode CPUexcludes CPU used at real-time and nice priorities.

APP_CPU_NORMAL_UTILThe percentage of time that processes in this group were in user moderunning at normal priority during the interval. Normal priority user modeCPU excludes CPU used at real-time and nice priorities.



55

APP_CPU_REALTIME_TIMEThe time, in seconds, that the processes in this group were in user modeat a “realtime” priority during the interval. “Realtime” priority is 0-127.

APP_CPU_REALTIME_UTILThe percentage of time that processes in this group were in user mode ata “realtime” priority during the interval. “Realtime” priority is 0-127.


APP_CPU_SYS_MODE_TIMEThe time, in seconds, during the interval that the CPU was in system modefor processes in this group.


APP_CPU_SYS_MODE_UTILThe percentage of time during the interval that the CPU was used insystem mode for processes in this group.


High system CPU utilizations are normal for IO intensive groups.Abnormally high system CPU utilization can indicate that a hardwareproblem is causing a high interrupt rate. It can also indicate programs thatare not making efficient system calls.

APP_CPU_TOTAL_TIMEThe total CPU time, in seconds, devoted to processes in this group duringthe interval.



56

APP_CPU_TOTAL_UTILThe percentage of the total CPU time devoted to processes in this groupduring the interval. This indicates the relative CPU load placed on thesystem by processes in this group.


Large values for this metric may indicate that this group is causing a CPUbottleneck. This would be normal in a computation-bound workload, butmight mean that processes are using excessive CPU time and perhapslooping.

If the “other” application shows significant amounts of CPU, you may wantto consider tuning your parm file so that process activity is accounted for inknown applications.

APP_CPU_TOTAL_UTIL =

APP_CPU_SYS_MODE_UTIL +

APP_CPU_USER_MODE_UTIL

APP_CPU_TOTAL_UTIL_CUMThe average CPU time per interval for processes in this group over thecumulative collection time, or since the last PRM configuration change.

The cumulative collection time is defined from the point in time wheneither: a) the process or kernel thread was first started, or b) theperformance tool was first started, or c) the cumulative counters werereset (relevant only to GlancePlus), whichever occurred last.

APP_CPU_USER_MODE_TIMEThe time, in seconds, that processes in this group were in user modeduring the interval. User CPU is the time spent in user mode at a normalpriority, at real-time priority, and at a nice priority.


APP_CPU_USER_MODE_UTILThe percentage of time that processes in this group were using the CPU inuser mode during the interval.

High user mode CPU percentages are normal for computation-intensivegroups. Low values of user CPU utilization compared to relatively highvalues for APP_CPU_SYS_MODE_UTIL can indicate a hardware problemor improperly tuned programs in this group.


57

User CPU is the time spent in user mode at a normal priority, at real-timepriority, and at a nice priority.


APP_DISK_FS_IO_RATEThe number of file system disk IOs for processes in this group during theinterval. Only local disks are counted in this measurement. NFS devicesare excluded.

These are physical IOs generated by user file system access and do notinclude virtual memory IOs, system IOs (inode updates), or IOs relating toraw disk access. An exception is user files accessed via the mmap(2) call,which will not show their physical IOs in this category. They appear undervirtual memory IOs.

APP_DISK_LOGL_IO_RATEThe number of logical IOs per second for processes in this group duringthe interval. Only local disks are counted in this measurement. NFSdevices are excluded.

Logical disk IOs are measured by counting the read and write systemcalls that are directed to disk devices. Also counted are read and writesystem calls made indirectly through other system calls, including readv,recvfrom, recv, recvmsg, ipcrecvcn, recfrom, writev, send, sento,sendmsg, and ipcsend.

There are several reasons why logical IOs may not correspond withphysical IOs. Logical IOs may not always result in a physical disk access,since the data may already reside in memory -- either in the buffer cache,or in virtual memory if the IO is to a memory mapped file. Several logicalIOs may all map to the same physical page or block. In these two cases,logical IOs are greater than physical IOs.

The reverse can also happen. A single logical write can cause a physicalread to fetch the block to be updated from disk, and then cause a physicalwrite to put it back on disk. A single logical IO can require more than onephysical page or block, and these can be found on different disks.Mirrored disks further distort the relationship between logical and physicalIO, since physical writes are doubled.

APP_DISK_LOGL_READThe number of logical reads for processes in this group during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.

Logical disk IOs are measured by counting the read system calls that aredirected to disk devices. Also counted are read system calls made


58

indirectly through other system calls, including readv, recvfrom, recv,recvmsg, ipcrecvcn, recfrom, send, sento, sendmsg, and ipcsend.



APP_DISK_LOGL_READ_RATEThe number of logical reads per second for processes in this group duringthe interval. Only local disks are counted in this measurement. NFSdevices are excluded.

Logical disk IOs are measured by counting the read system calls that aredirected to disk devices. Also counted are read system calls madeindirectly through other system calls, including readv, recvfrom, recv,recvmsg, ipcrecvcn, recfrom, send, sento, sendmsg, and ipcsend.



APP_DISK_LOGL_WRITEThe number of logical writes for processes in this group during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.

Logical disk IOs are measured by counting the write system calls that aredirected to disk devices. Also counted are write system calls madeindirectly through other system calls, including writev, recvfrom, recv,recvmsg, ipcrecvcn, recfrom, send, sento, sendmsg, and ipcsend.

There are several reasons why logical IOs may not correspond withphysical IOs. Logical IOs may not always result in a physical disk access,since the data may already reside in memory -- either in the buffer cache,


59

or in virtual memory if the IO is to a memory mapped file. Several logicalIOs may all map to the same physical page or block. In these two cases,logical IOs are greater than physical IOs.


APP_DISK_LOGL_WRITE_RATEThe number of logical writes per second for processes in this group duringthe interval. Only local disks are counted in this measurement. NFSdevices are excluded.




APP_DISK_PHYS_IO_RATEThe number of physical IOs per second for processes in this group duringthe interval.

APP_DISK_PHYS_READThe number of physical reads for processes in this group during theinterval.

APP_DISK_PHYS_READ_RATEThe number of physical reads per second for processes in this groupduring the interval.


60

APP_DISK_PHYS_WRITEThe number of physical writes for processes in this group during theinterval.

APP_DISK_PHYS_WRITE_RATEThe number of physical writes per second for processes in this groupduring the interval.

APP_DISK_RAW_IO_RATEThe total number of raw IOs for processes in this group during the interval.Only accesses to local disk devices are counted.

APP_DISK_SUBSYSTEM_QUEUEThe average number of processes or kernel threads in this group thatwere blocked on the disk subsystem (waiting for their file system IOs tocomplete) during the interval.

This is the sum of processes or kernel threads in the DISK, INODE,CACHE and CDFS wait states. It does not include processes or kernelthreads doing raw IO to disk devices.

The Application QUEUE metrics, which are based on block states,represent the average number of process or kernel thread counts, notactual queues, within the context of a specific application.

The Application WAIT PCT metrics, which are also based on block states,represent the percentage of processes or kernel threads that were alive onthe system within the context of a specific application. These values willvary greatly depending on the application.

No direct comparison is reasonable with the Global Queue metrics sincethey represent the average number of all processes or kernel threads thatwere alive on the system. As such, the Application WAIT PCT metricscannot be summed or compared with global values easily. In addition, thesum of each Application WAIT PCT for all applications will not equal 100%since these values will vary greatly depending on the number of processesor kernel threads in each application.

For example, the GBL_DISK_SUBSYSTEM_QUEUE values can be low,while the APP_DISK_SUBSYSTEM_WAIT_PCT values can be high. Inthis case, there are many processes on the system, but there are only avery small number of processes in the specific application that is beingexamined and there is a high percentage of those few processes that areblocked on the disk I/O subsystem.


61

APP_DISK_SUBSYSTEM_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on the disk subsystem (waiting for their file system IOs tocomplete) during the interval.

This is the sum of processes or kernel threads in the DISK, INODE,CACHE and CDFS wait states. It does not include processes or kernelthreads doing raw IO to disk devices.

A percentage of time spent in a wait state is calculated as theaccumulated time kernel threads belonging to processes in this groupspent waiting in this state, divided by accumulated alive time of kernelthreads belonging to processes in this group during the interval.

For example, assume an application has 20 kernel threads. During theinterval, ten kernel threads slept the entire time, while ten kernel threadswaited on terminal input. As a result, the application wait percent valueswould be 50% for SLEEP and 50% for TERM (that is, terminal IO).





APP_DISK_SYSTEM_IO_RATEThe number of physical IOs per second generated by the kernel for filesystem management (inode accesses or updates) for processes in thisgroup during the interval.

APP_DISK_VM_IO_RATEThe number of virtual memory IOs per second made on behalf ofprocesses in this group during the interval.

IOs to user file data are not included in this metric unless they were donevia the mmap(2) system call.


62

APP_INTERVALThe amount of time in the interval.

APP_INTERVAL_CUMThe amount of time over the cumulative collection time.


APP_IO_BYTEThe number of characters (in KB) transferred for processes in this group toall devices during the interval. This includes IO to disk, terminal, tape andprinters.

APP_IO_BYTE_RATEThe number of characters (in KB) per second transferred for processes inthis group to all devices during the interval. This includes IO to disk,terminal, tape and printers.

APP_IPC_SUBSYSTEM_QUEUEThe average number of processes or kernel threads in this group blockedon the InterProcess Communication (IPC) subsystems (waiting for theirinterprocess communication activity to complete) during the interval.

This is the sum of processes or kernel threads in the IPC, MSG, SEM,PIPE, SOCKT (that is, sockets) and STRMS (that is, streams IO) waitstates.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (IPC + MSG + SEM + PIPE + SOCKT +STRMS) divided by the interval time.



No direct comparison is reasonable with the Global Queue metrics sincethey represent the average number of all processes or kernel threads thatwere alive on the system. As such, the Application WAIT PCT metricscannot be summed or compared with global values easily. In addition, the


63

sum of each Application WAIT PCT for all applications will not equal 100%since these values will vary greatly depending on the number of processesor kernel threads in each application.


APP_IPC_SUBSYSTEM_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on the InterProcess Communication (IPC) subsystems (waiting fortheir interprocess communication activity to complete) during the interval.

This is the sum of processes or kernel threads in the IPC, MSG, SEM,PIPE, SOCKT (that is, sockets) and STRMS (that is, streams IO) waitstates.








64

APP_MAJOR_FAULTThe number of major page faults that required a disk IO for processes inthis group during the interval.

APP_MAJOR_FAULT_RATEThe number of major page faults per second that required a disk IO forprocesses in this group during the interval.

APP_MEM_QUEUEThe average number of processes or kernel threads in this group blockedon memory (waiting for virtual memory disk accesses to complete) duringthe interval.

This typically happens when processes or kernel threads are allocating alarge amount of memory. It can also happen when processes or kernelthreads access memory that has been paged out to disk (deactivated)because of overall memory pressure on the system. Note that largeprograms can block on VM disk access when they are initializing, bringingtheir text and data pages into memory.





APP_MEM_RESThe size (in KB) of resident memory for processes in this group that werealive at the end of the interval. This consists of text, data, stack, as well asthe process' portion of shared memory regions (such as, shared libraries,text segments, and shared data).

For each process, resident memory (RSS) is calculated as


65

RSS = sum of private region pages +

(sum of shared region pages /

number of references)

The number of references is a count of the number of attachments to thememory region. Attachments, for shared regions, may come from severalprocesses sharing the same memory, a single process with multipleattachments, or combinations of these.

This value is only updated when a process uses CPU. Thus, undermemory pressure, this value may be higher than the actual amount ofresident memory for processes which are idle.

Refer to the help text for PROC_MEM_RES for additional information.

APP_MEM_UTILThe approximate percentage of the system's physical memory used asresident memory by processes in this group that were alive at the end ofthe interval.

This metric summarizes process private and shared memory in eachapplication. This consists of text, data, stack, as well the process' portionof shared memory regions (such as, shared libraries, text segments, andshared data). The sum of the shared region pages is divided by thenumber of references.

Each application's total resident memory is summed. This value is thendivided by the summed total of all applications resident memory and thenmultiplied by the ratio of available user memory versus total physicalmemory to arrive at a computed percent of total physical memory.

It must be remembered, however, that this is a computed metric thatshows the approximate percentage of the physical memory used asresident memory by the processes in this application during the interval.

This metric is not available for HP-UX OV PerformanceAgent. It isavailable for HP-UX GlancePlus.

APP_MEM_VIRTThe approximate size (in KB) of virtual memory for processes in this groupthat were alive at the end of the interval.

This is the sum of the virtual memory region sizes for all processes in thisgroup. Since this virtual memory size for each process includes sharedregions, such as library text and data, the shared regions are countedmultiple times in this metric. For example, if two processes are attached toa 10MB shared region, then 20MB is reported in this metric.

This value is not affected by the reference count. As such, this metric canoverestimate the virtual memory being used by processes in this groupwhen they share memory regions.


66

APP_MEM_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on memory (waiting for virtual memory disk accesses to complete)during the interval.







APP_MINOR_FAULTThe number of minor page faults satisfied in memory (a page wasreclaimed from one of the free lists) for processes in this group during theinterval.

APP_MINOR_FAULT_RATEThe number of minor page faults per second satisfied in memory (pageswere reclaimed from one of the free lists) for processes in this groupduring the interval.


67

APP_NAMEThe name of the application (up to 20 characters). This comes from theparm file where the applications are defined.

The application called “other” captures all other processes not defined inthe parm file.

APP_NAME_PRM_GROUPNAMEThe PRM group name. The PRM group configuration is kept in the PRMconfiguration file.

APP_NETWORK_SUBSYSTEM_QUEUEThe average number of processes or kernel threads in this group wereblocked on the network subsystem (waiting for their network activity tocomplete) during the interval.

This is the sum of processes or kernel threads in the LAN, NFS, and RPCwait states. This does not include processes or kernel threads blocked onSOCKT (that is, socket) waits, as some processes or kernel threads sitidle in SOCKT waits for long periods.





APP_NETWORK_SUBSYSTEM_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on the network subsystem (waiting for their network activity tocomplete) during the interval.


68

This is the sum of processes or kernel threads in the LAN, NFS, and RPCwait states. This does not include processes or kernel threads blocked onSOCKT (that is, socket) waits, as some processes or kernel threads sitidle in SOCKT waits for long periods.

This is calculated as the accumulated time that all processes or kernelthreads in this group spent blocked on (LAN + NFS + RPC) divided by theinterval time.







APP_NUMThe sequentially assigned number of this application.

APP_OTHER_IO_QUEUEThe average number of processes or kernel threads in this group thatwere blocked on “other IO” during the interval. “Other IO” includes all IOdirected at a device (connected to the local computer) which is not aterminal or LAN. Examples of “other IO” devices are local printers, tapes,instruments, and disks. Time waiting for character (raw) IO to disks isincluded in this measurement. Time waiting for file system buffered IO to


69

disks will typically been seen as IO or CACHE wait. Time waiting for IO toNFS disks is reported as NFS wait.

This is calculated as the accumulated time that all processes or kernelthreads in this group spent blocked on other IO divided by the intervaltime.





APP_OTHER_IO_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on “other IO” during the interval. “Other IO” includes all IOdirected at a device (connected to the local computer) which is not aterminal or LAN. Examples of “other IO” devices are local printers, tapes,instruments, and disks. Time waiting for character (raw) IO to disks isincluded in this measurement. Time waiting for file system buffered IO todisks will typically been seen as IO or CACHE wait. Time waiting for IO toNFS disks is reported as NFS wait.




The Application WAIT PCT metrics, which are also based on block states,represent the percentage of processes or kernel threads that were alive on


70

the system within the context of a specific application. These values willvary greatly depending on the application.



APP_PRIThe average priority of the processes in this group during the interval.

APP_PRI_QUEUEThe average number of processes or kernel threads in this group blockedon PRI (waiting for their priority to become high enough to get the CPU)during the interval.

This is calculated as the accumulated time that all processes or kernelthreads in this group spent blocked on PRI divided by the interval time.






71

APP_PRI_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on PRI (waiting for their priority to become high enough to get theCPU) during the interval.







APP_PRM_CPUCAP_MODEThe PRM CPU Cap Mode state on this system:

0 = PRM is not installed or not

configured.

1 = CPU Cap Mode is not enabled

(PRM CPU entitlements are

in effect)

2 = CPU Cap Mode is enabled

(The PRM CPU entitlements

behave as caps or limits)


72

APP_PRM_CPU_ENTITLEMENTThe PRM CPU entitlement for this PRM Group ID entry as defined in thePRM configuration file.

APP_PRM_CPU_TOTAL_UTIL_CUMThe average CPU time per interval for processes in this group over thecumulative collection time, or since the last PRM configuration change.


APP_PRM_DISK_STATEThe PRM DISK state on this system:

0 = PRM is not installed or no

disk specification

1 = reset (PRM is installed in

reset condition or no disk

specification)

2 = configured/disabled (The PRM

disk management is configured)

3 = enabled/configured (The PRM

disk management is enabled and

volume groups are configured)

4 = enabled/unconfigured (The PRM

disk management is enabled,

however, no volume groups are

configured)

APP_PRM_GROUPIDThe PRM Group ID. The PRM group configuration is kept in the PRMconfiguration file.

APP_PRM_INTERVAL_CUMThe amount of time over the cumulative collection time.



73

APP_PRM_MEM_AVAILPRM available memory is the amount of physical memory less the amountof memory reserved for the kernel and system processes running in thePRM_SYS group 0. PRM available memory is a dynamic value thatchanges with system usage.

This metric is available on HP-UX 11.0 and beyond.

APP_PRM_MEM_ENTITLEMENTThe PRM MEM entitlement for this PRM Group ID entry as defined in thePRM configuration file.

APP_PRM_MEM_STATEThe PRM MEM state on this system:

0 = PRM is not installed or no

memory specification

1 = reset (PRM is installed in

reset condition or no memory

specification)

2 = configured/disabled (The PRM

memory scheduler is configured,

but the standard HP-UX

scheduler is in effect)

3 = enabled (The PRM memory

scheduler is configured and

in effect)

APP_PRM_MEM_UPPERBOUNDThe PRM MEM upperbound for this PRM Group ID entry as defined in thePRM configuration file.

APP_PRM_MEM_UTILThe percent of PRM memory used by processes (process private spaceplus a process' portion of shared memory) within the PRM groups duringthe interval.

PRM available memory is the amount of physical memory less the amountof memory reserved for the kernel and system processes running in thePRM_SYS group 0. PRM available memory is a dynamic value thatchanges with system usage.


74

APP_PRM_STATEThe PRM CPU state on this system:

0 = PRM is not installed

1 = reset (PRM is configured with

only the system group. The

standard HP-UX CPU scheduler

is in effect)

2 = configured/disabled (the PRM

CPU scheduler is configured,

but the standard HP-UX

scheduler is in effect)

3 = enabled (the PRM CPU scheduler

is configured and in effect)

APP_PRM_SUSPENDED_PROCThe number of processes within the PRM groups that were suspendedduring the interval.

APP_PROC_RUN_TIMEThe average run time for processes in this group that completed during theinterval.

APP_SAMPLEThe number of samples of process data that have been averaged oraccumulated during this sample.

APP_SEM_QUEUEThe average number of processes or kernel threads in this group thatwere blocked on semaphores (waiting for their semaphore operations tocomplete) during the interval.



No direct comparison is reasonable with the Global Queue metrics sincethey represent the average number of all processes or kernel threads thatwere alive on the system. As such, the Application WAIT PCT metricscannot be summed or compared with global values easily. In addition, thesum of each Application WAIT PCT for all applications will not equal 100%


75

since these values will vary greatly depending on the number of processesor kernel threads in each application.


APP_SEM_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on semaphores (waiting for their semaphore operations tocomplete) during the interval.







APP_SLEEP_QUEUEThe average number of processes or kernel threads in this group thatwere blocked on SLEEP (waiting to awaken from sleep system calls)during the interval. A process or kernel thread enters the SLEEP state by


76

putting itself to sleep using system calls such as sleep, wait, pause,sigpause, sigsuspend, poll and select.





APP_SLEEP_WAIT_PCTThe percentage of time processes or kernel threads in this group wereblocked on SLEEP (waiting to awaken from sleep system calls) during theinterval. A process or kernel thread enters the SLEEP state by puttingitself to sleep using system calls such as sleep, wait, pause, sigpause,sigsuspend, poll and select.





No direct comparison is reasonable with the Global Queue metrics sincethey represent the average number of all processes or kernel threads thatwere alive on the system. As such, the Application WAIT PCT metricscannot be summed or compared with global values easily. In addition, the


77

sum of each Application WAIT PCT for all applications will not equal 100%since these values will vary greatly depending on the number of processesor kernel threads in each application.


APP_TIMEThe end time of the measurement interval.

BYCPU_CPU_CSWITCH_TIMEThe time, in seconds, that this CPU was performing context switchesduring the interval. This includes context switches that result in theexecution of a different process and those caused by a process stopping,then resuming, with no other process running in the meantime.

BYCPU_CPU_CSWITCH_TIME_CUMThe time, in seconds, that this CPU was performing context switches overthe cumulative collection time. This includes context switches that result inthe execution of a different process and those caused by a processstopping, then resuming, with no other process running in the meantime.


BYCPU_CPU_CSWITCH_UTILThe percentage of time that this CPU was performing context switchesduring the interval. This includes context switches that result in theexecution of a different process and those caused by a process stopping,then resuming, with no other process running in the meantime.

BYCPU_CPU_CSWITCH_UTIL_CUMThe percentage of time that this CPU was performing context switchesover the cumulative collection time. This includes context switches thatresult in the execution of a different process and those caused by aprocess stopping, then resuming, with no other process running in themeantime.

The cumulative collection time is defined from the point in time wheneither: a) the process or kernel thread was first started, or b) the


78

performance tool was first started, or c) the cumulative counters werereset (relevant only to GlancePlus), whichever occurred last.

BYCPU_CPU_INTERRUPT_TIMEThe time, in seconds, that this CPU was performing interrupt processingduring the interval.

BYCPU_CPU_INTERRUPT_TIME_CUMThe time, in seconds, that this CPU was performing interrupt processingover the cumulative collection time.


BYCPU_CPU_INTERRUPT_UTILThe percentage of time that this CPU was performing interrupt processingduring the interval.

BYCPU_CPU_INTERRUPT_UTIL_CUMThe percentage of time that this CPU was performing interrupt processingover the cumulative collection time.


BYCPU_CPU_NICE_TIMEThe time, in seconds, that this CPU was in user mode at a nice priorityduring the interval. The NICE metrics include positive nice value CPUtime only. Negative nice value CPU is broken out into NNICE (negativenice) metrics. Positive nice values range from 20 to 39. Negative nicevalues range from 0 to 19.

BYCPU_CPU_NICE_TIME_CUMThe time, in seconds, that this CPU was in user mode at a nice priorityover the cumulative collection time. The NICE metrics include positivenice value CPU time only. Negative nice value CPU is broken out intoNNICE (negative nice) metrics. Positive nice values range from 20 to 39.Negative nice values range from 0 to 19.


79


BYCPU_CPU_NICE_UTILThe percentage of time that this CPU was in user mode at a nice priorityduring the interval. The NICE metrics include positive nice value CPUtime only. Negative nice value CPU is broken out into NNICE (negativenice) metrics. Positive nice values range from 20 to 39. Negative nicevalues range from 0 to 19.

BYCPU_CPU_NICE_UTIL_CUMThe average percentage of time that this CPU was in user mode at a nicepriority over the cumulative collection time. The NICE metrics includepositive nice value CPU time only. Negative nice value CPU is broken outinto NNICE (negative nice) metrics. Positive nice values range from 20 to39. Negative nice values range from 0 to 19.


BYCPU_CPU_NNICE_TIMEThe time, in seconds, that this CPU was in user mode at a nice prioritycomputed from processes with negative nice values during the interval.

BYCPU_CPU_NNICE_TIME_CUMThe time, in seconds, that this CPU was in user mode at a nice prioritycomputed from processes with negative nice values over the cumulativecollection time.


BYCPU_CPU_NNICE_UTILThe percentage of time that this CPU was in user mode at a nice prioritycomputed from processes with negative nice values during the interval.


80

BYCPU_CPU_NNICE_UTIL_CUMThe average percentage of time that this CPU was in user mode at a nicepriority computed from processes with negative nice values over thecumulative collection time.


BYCPU_CPU_NORMAL_TIMEThe time, in seconds, that this CPU was running in user mode at a normalpriority during the interval. Normal priority user mode CPU excludes CPUused at real-time and nice priorities.

BYCPU_CPU_NORMAL_TIME_CUMThe time, in seconds, that this CPU was running in user mode at a normalpriority over the cumulative collection time. Normal priority user modeCPU excludes CPU used at real-time and nice priorities.


BYCPU_CPU_NORMAL_UTILThe percentage of time that this CPU was running in user mode at anormal priority during the interval. Normal priority user mode CPUexcludes CPU used at real-time and nice priorities.

BYCPU_CPU_NORMAL_UTIL_CUMThe average percentage of time that this CPU was running in user modeat a normal priority over the cumulative collection time. Normal priorityuser mode CPU excludes CPU used at real-time and nice priorities.


BYCPU_CPU_REALTIME_TIMEThe time, in seconds, that this CPU was running at a realtime priorityduring the interval.


81

BYCPU_CPU_REALTIME_TIME_CUMThe time, in seconds, that this CPU was running at a realtime priority overthe cumulative collection time.


BYCPU_CPU_REALTIME_UTILThe percentage of time that this CPU was running at a realtime priorityduring the interval.

BYCPU_CPU_REALTIME_UTIL_CUMThe percentage of time that this CPU was running at a realtime priorityover the cumulative collection time.


BYCPU_CPU_SYSCALL_TIMEThe time, in seconds, that this CPU was running in system mode (notincluding interrupt, context switch, trap or vfault CPU) during the lastinterval.

BYCPU_CPU_SYSCALL_TIME_CUMThe time, in seconds, that this CPU was running in system mode (notincluding interrupt, context switch, trap or vfault CPU) over the cumulativecollection time.


BYCPU_CPU_SYSCALL_UTILThe percentage of time that this CPU was running in system mode (notincluding interrupt, context switch, trap or vfault CPU) during the interval.


82

BYCPU_CPU_SYSCALL_UTIL_CUMThe average percentage of time that this CPU was running in systemmode (not including interrupt, context switch, trap or vfault CPU) over thecumulative collection time.


BYCPU_CPU_SYS_MODE_TIMEThe time, in seconds, that this CPU was in system mode during theinterval.

BYCPU_CPU_SYS_MODE_TIME_CUMThe time, in seconds, that this CPU was in system mode over thecumulative collection time.


BYCPU_CPU_SYS_MODE_UTILThe percentage of time that this CPU was in system mode during theinterval.

BYCPU_CPU_SYS_MODE_UTIL_CUMThe percentage of time that this CPU was in system mode over thecumulative collection time.


BYCPU_CPU_TOTAL_UTILThe percentage of time that this CPU was not idle during the interval.

BYCPU_CPU_TOTAL_UTIL_CUMThe average percentage of time that this CPU was not idle over thecumulative collection time.


83


BYCPU_CPU_TRAP_TIMEThe time, in seconds, this CPU was in trap handler code during theinterval.

BYCPU_CPU_TRAP_TIME_CUMThe time, in seconds, this CPU was in trap handler code over thecumulative collection time.


BYCPU_CPU_TRAP_UTILThe percentage of time this CPU was in trap handler code during theinterval.

BYCPU_CPU_TRAP_UTIL_CUMThe average percentage of time this CPU was in trap handler code overthe cumulative collection time.


BYCPU_CPU_USER_MODE_TIMEThe time, in seconds, during the interval that this CPU was in user mode.User CPU is the time spent in user mode at a normal priority, at real-timepriority, and at a nice priority.

BYCPU_CPU_USER_MODE_TIME_CUMThe time, in seconds, that this CPU was in user mode over the cumulativecollection time. User CPU is the time spent in user mode at a normalpriority, at real-time priority, and at a nice priority.



84


BYCPU_CPU_USER_MODE_UTILThe percentage of time that this CPU was in user mode during the interval.User CPU is the time spent in user mode at a normal priority, at real-timepriority, and at a nice priority.

BYCPU_CPU_USER_MODE_UTIL_CUMThe average percentage of time that this CPU was in user mode over thecumulative collection time. User CPU is the time spent in user mode at anormal priority, at real-time priority, and at a nice priority.


BYCPU_CPU_VFAULT_TIMEThe time, in seconds, this CPU was handling page faults during theinterval.

BYCPU_CPU_VFAULT_TIME_CUMThe time, in seconds, this CPU was handling page faults over thecumulative collection time.


BYCPU_CPU_VFAULT_UTILThe percentage of time this CPU was handling page faults during theinterval.

BYCPU_CPU_VFAULT_UTIL_CUMThe average percentage of time this CPU was handling page faults overthe cumulative collection time.



85

BYCPU_CSWITCHThe number of context switches for this CPU during the interval.

This includes context switches that result in the execution of a differentprocess and those caused by a process stopping, then resuming, with noother process running in the meantime.

BYCPU_CSWITCH_CUMThe number of context switches for this CPU over the cumulativecollection time.



BYCPU_CSWITCH_RATEThe average number of context switches per second for this CPU duringthe interval.


BYCPU_CSWITCH_RATE_CUMThe average number of context switches per second for this CPU over thecumulative collection time.



BYCPU_IDThe ID number of this CPU.


86

BYCPU_INTERRUPT_RATEThe average number of device interrupts per second for this CPU duringthe interval. A value of “na” is displayed on a system with multiple CPUs.

BYCPU_INTERRUPT_STATEA text string indicating whether the current processor is “enabled” or“disabled” for servicing IO interrupts.

BYCPU_LAST_PROC_IDThe process id (pid) of the last process to have used this CPU.

BYCPU_LAST_THREAD_IDThe thread ID (TID) number of the last kernel thread to have used thisCPU.


BYCPU_LAST_USER_THREAD_IDThe user thread ID number of the last user thread to have used this CPUwithin the context of its associated process. A process may have multipleuser threads. This indicates the most recently executed user thread of theprocess identified in BYCPU_LAST_PROC_ID.


BYCPU_RUN_QUEUE_15_MINThis represents the 15 minute load average for this processor.

BYCPU_RUN_QUEUE_1_MINThis represents the 1 minute load average for this processor. This metricis derived from a kernel variable (avenrun) which is calculated by summingthe number of runnable processes or kernel threads for each processorand averaging the samples over the last minute. Processes or kernelthreads marked “runnable” include:

* a process or kernel thread using

the CPU at the time of the sample

* a process or kernel thread waiting

for the CPU at the time of the

sample

* a process or kernel thread paused


87

on a “short disk wait” at the time

of the sample (HP-UX 10.20 and

11.0)

On HP-UX 10.20 and 11.0, this metric can include processes or kernelthreads which are waiting for disk IO to complete. Because of that, it is nota reliable CPU bottleneck indicator.

On HP-UX 11i, this metric does not include processes or kernel threadswhich are waiting for disk IO to complete.

Several standard UNIX commands, such as uptime(1), display avenrun asthe “1-minute Load Average.”

BYCPU_RUN_QUEUE_5_MINThis represents the 5 minute load average for this processor.

BYCPU_STATEA text string indicating the current state of a processor is “enabled”,“disabled” or “unknown”.

BYDSKDETAIL_LABELThe type of entry this disk device is associated with - could be a partition,file system directory, logical volume, or volume group.

BYDSKDETAIL_NAMEThe name of the partition, file system directory, logical volume, or volumegroup this disk device is associated with.

BYDSK_AVG_SERVICE_TIMEThe average time, in milliseconds, that this disk device spent processingeach disk request during the interval. For example, a value of 5.14 wouldindicate that disk requests during the last interval took on average slightlylonger than five one-thousandths of a second to complete for this device.

This is a measure of the speed of the disk, because slower disk devicestypically show a larger average service time. Average service time is alsodependent on factors such as the distribution of I/O requests over theinterval and their locality. It can also be influenced by disk driver andcontroller features such as I/O merging and command queueing. Notethat this service time is measured from the perspective of the kernel, notthe disk device itself. For example, if a disk device can find the requesteddata in its cache, the average service time could be quicker than thespeed of the physical disk hardware.

This metric can be used to help determine which disk devices are takingmore time than usual to process requests.


88

BYDSK_BUSThe name of the bus interface used by this disk.

BYDSK_CONTROLLERThe disk controller name. This information is only available for disks usingthe hpib or hpfl interfaces.

BYDSK_DEVNAMEThe name identifying the specific disk spindle is the hardware path whichspecifies the address of the hardware components leading to the diskdevice.

BYDSK_DEVNOMajor / Minor number of the device.

BYDSK_DIRNAMEThe name of the file system directory mounted on this disk device. If morethan one file system is mounted on this device, “Multiple FS” is seen.

BYDSK_FS_IO_RATEThe number of physical file system reads and writes per second to thisdisk device during the interval.

BYDSK_FS_READThe number of physical file system reads from this disk device during theinterval.

BYDSK_FS_READ_RATEThe number of physical file system reads per second from this disk deviceduring the interval.

BYDSK_FS_WRITEThe number of physical file system writes to this disk device during theinterval.


89

BYDSK_FS_WRITE_RATEThe number of physical file system writes per second to this disk deviceduring the interval.

BYDSK_INTERVALThe amount of time in the interval.

BYDSK_INTERVAL_CUMThe amount of time over the cumulative collection time.


BYDSK_LOGL_BYTE_RATEThe number of logical read or write KBs per second to this disk deviceduring the interval.

Logical IO rates by disk device cannot be obtained in a multi-disk LVMconfiguration because there is no reasonable means of tying logical IOtransactions to physical spindles spanned on the logical volume.

Therefore, if you have a multi-disk LVM configuration, you always see “na”for this metric.

BYDSK_LOGL_BYTE_RATE_CUMThe average number of KBs of logical read or writes to this disk deviceover the cumulative collection time.




BYDSK_LOGL_IO_RATEThe total number of logical IOs per second for this disk device during theinterval.


90



BYDSK_LOGL_IO_RATE_CUMThe average number of logical IOs per second over the cumulativecollection time.




BYDSK_LOGL_READThe number of logical reads for this disk device during the interval.



BYDSK_LOGL_READ_BYTE_RATEThe number of logical read KBs per second from this disk device duringthe interval.



BYDSK_LOGL_READ_BYTE_RATE_CUMThe average number of logical KBs per second read from this disk deviceover the cumulative collection time.



91




BYDSK_LOGL_READ_RATEThe number of logical reads per second for this disk device during theinterval.



BYDSK_LOGL_READ_RATE_CUMThe average number of logical reads per second for this disk device overthe cumulative collection time.




BYDSK_LOGL_WRITEThe number of logical writes for this disk device during the interval.



BYDSK_LOGL_WRITE_BYTE_RATEThe number of logical writes KBs per second to this disk device during theinterval.


92



BYDSK_LOGL_WRITE_BYTE_RATE_CUMThe average number of KBs of logical writes per second to this disk deviceover the cumulative collection time.




BYDSK_LOGL_WRITE_RATEThe number of logical writes per second for this disk device during theinterval.



BYDSK_LOGL_WRITE_RATE_CUMThe average number of logical writes per second for this disk device overthe cumulative collection time.





93

BYDSK_PHYS_BYTE_RATEThe average KBs per second transferred to or from this disk device duringthe interval.

BYDSK_PHYS_BYTE_RATE_CUMThe average number of KBs per second of physical reads and writes tothis disk device over the cumulative collection time.


BYDSK_PHYS_IOThe number of physical IOs for this disk device during the interval.

BYDSK_PHYS_IO_RATEThe average number of physical IO requests per second for this diskdevice during the interval. This counts disk reads and writes of all types,including virtual memory and raw IO.

BYDSK_PHYS_IO_RATE_CUMThe average number of physical reads and writes per second for this diskdevice over the cumulative collection time.


BYDSK_PHYS_READThe number of physical reads for this disk device during the interval.

BYDSK_PHYS_READ_BYTEThe KBs transferred from this disk device during the interval. This countsall types of disk reads, including file system, virtual memory, and raw IO.


94

BYDSK_PHYS_READ_BYTE_RATEThe average KBs per second transferred from this disk device during theinterval. This counts all types of disk reads, including file system, virtualmemory, and raw IO.

BYDSK_PHYS_READ_BYTE_RATE_CUMThe average number of KBs per second of physical reads from this diskdevice over the cumulative collection time.


BYDSK_PHYS_READ_RATEThe average number of physical reads per second for this disk deviceduring the interval.

BYDSK_PHYS_READ_RATE_CUMThe average number of physical reads per second for this disk device overthe cumulative collection time.


BYDSK_PHYS_WRITEThe number of physical writes for this disk device during the interval.

BYDSK_PHYS_WRITE_BYTEThe KBs transferred to this disk device during the interval. This counts alltypes of disk writes, including file system, virtual memory, and raw IO.

BYDSK_PHYS_WRITE_BYTE_RATEThe average KBs per second transferred to this disk device during theinterval. This counts all types of writes, including file system, virtualmemory, and raw IO.


95

BYDSK_PHYS_WRITE_BYTE_RATE_CUMThe average number of KBs per second of physical writes to this diskdevice over the cumulative collection time.


BYDSK_PHYS_WRITE_RATEThe average number of physical writes per second for this disk deviceduring the interval.

BYDSK_PHYS_WRITE_RATE_CUMThe average number of physical writes per second for this disk deviceover the cumulative collection time.


BYDSK_PRODUCT_IDThe disk product ID.

BYDSK_QUEUE_0_UTILThe percentage of intervals during which there were no IO requestspending for this disk device over the cumulative collection time.


For example if 4 intervals have passed (that is, 4 screen updates) and theaverage queue length for these intervals was 0, 1.5, 0, and 3, then thevalue for this metric would be 50% since 50% of the intervals had a zeroqueue length.

BYDSK_QUEUE_2_UTILThe percentage of intervals during which there were 1 or 2 IO requestspending for this disk device over the cumulative collection time.



96


For example if 4 intervals have passed (that is, 4 screen updates) and theaverage queue length for these intervals was 0, 1, 0, and 2, then the valuefor this metric would be 50% since 50% of the intervals had a 1-2 queuelength.

BYDSK_QUEUE_4_UTILThe percentage of intervals during which there were 3 or 4 IO requestswaiting to use this disk device over the cumulative collection time.



BYDSK_QUEUE_8_UTILThe percentage of intervals during which there were between 5 and 8 IOrequests pending for this disk device over the cumulative collection time.



BYDSK_QUEUE_X_UTILThe percentage of intervals during which there were more than 8 IOrequests pending for this disk device over the cumulative collection time.


For example if 4 intervals have passed (that is, 4 screen updates) and theaverage queue length for these intervals was 0, 9, 0, and 10, then thevalue for this metric would be 50% since 50% of the intervals had queuelength greater than 8.


97

BYDSK_RAW_IO_RATEThe number of raw reads or writes per second made to this disk deviceduring the interval.

BYDSK_RAW_READThe number of physical raw reads made from this disk device during theinterval.

BYDSK_RAW_READ_RATEThe number of raw reads per second made from this disk device duringthe interval.

BYDSK_RAW_WRITEThe number of physical raw writes made to this disk device during theinterval.

BYDSK_RAW_WRITE_RATEThe number of raw writes per second made to this disk device during theinterval.

BYDSK_REQUEST_QUEUEThe average number of IO requests that were in the wait queue for thisdisk device during the interval. These requests are the physical requests(as opposed to logical IO requests).

BYDSK_SYSTEM_IOThe number of physical system reads or writes to this disk device duringthe interval.

BYDSK_SYSTEM_IO_RATEThe number of physical system reads or writes per second to this diskdevice during the interval.

BYDSK_SYSTEM_READ_RATEThe number of physical system reads per second from this disk deviceduring the interval.


98

BYDSK_SYSTEM_WRITE_RATEThe number of physical system writes per second to this disk deviceduring the interval.

BYDSK_TIMEThe time of day of the interval.

BYDSK_UTILThe percentage of the time during the interval that the disk device had IOin progress from the point of view of the Operating System. In otherwords, the utilization or percentage of time busy servicing requests for thisdevice.

This is a measure of the ability of the IO path to meet the transferdemands being placed on it. Slower disk devices may show a higherutilization with lower IO rates than faster disk devices such as disk arrays.A value of greater than 50% utilization over time may indicate that thisdevice or its IO path is a bottleneck, and the access pattern of theworkload, database, or files may need reorganizing for better balance ofdisk IO load.

BYDSK_UTIL_CUMThe percentage of the time over the cumulative collection time that thedisk device had IO in progress from the point of view of the OperatingSystem. In other words, the utilization or percentage of time busyservicing requests for this device.


This is a measure of the ability of the IO path to meet the transferdemands being placed on it. Slower disk devices may show a higherutilization with lower IO rates than faster disk devices such as disk arrays.A value of greater than 50% utilization over time may indicate that thisdevice or its IO path is a bottleneck, and the access pattern of theworkload, database, or files may need reorganizing for better balance ofdisk IO load.

BYDSK_VENDOR_IDThe disk vendor ID. This information is only available for disks using thescsi interface.


99

BYDSK_VM_IOThe number of virtual memory IOs to this disk device during the interval.

BYDSK_VM_IO_RATEThe number of virtual memory IOs per second to this disk device duringthe interval.

BYDSK_VM_READ_RATEThe number of virtual memory reads per second from this disk deviceduring the interval.

BYDSK_VM_WRITE_RATEThe number of virtual memory writes per second to this disk device duringthe interval.

BYNETIF_COLLISIONThe number of physical collisions that occurred on the network interfaceduring the interval. A rising rate of collisions versus outbound packets isan indication that the network is becoming increasingly congested. Thismetric does not currently include deferred packets.

Physical statistics are packets recorded by the network drivers. Thesenumbers most likely will not be the same as the logical statistics. On HP-UX 11.0 and beyond, the values returned for the loopback interface willshow "na" for the physical statistics since there is no network driveractivity.

Logical statistics are packets seen only by the Interface Protocol (IP) layerof the networking subsystem. Not all packets seen by IP will go out andcome in through a network driver. Examples cases are the 127.0.0.1(loopback interface). Pings or other network generating commands (ftp,rlogin, and so forth) to 127.0.0.1 will not change physical driver statistics.Pings to IP addresses on remote systems will change physical driverstatistics.

This is different from pre-11.0 systems where commands addressed to thelocal host always went down to the driver and the logical and physicalcounters were always updated.

For HP-UX 10.20 and earlier releases, this is the same as the “Coll”column from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Single Collision Frames”, “Multiple Collision Frames”, “LateCollisions”, and “Excessive Collisions” values from the output of the“lanadmin” utility for the network interface. Remember that “lanadmin”


100

reports cumulative counts. For this release and beyond, “netstat -i” showsnetwork activity on the logical level (IP) only.

This metric is updated at the sampling interval, regardless of the numberof IP addresses on the system.

BYNETIF_COLLISION_1_MIN_RATEThe number of physical collisions per minute on the network interfaceduring the interval. A rising rate of collisions versus outbound packets isan indication that the network is becoming increasingly congested. Thismetric does not currently include deferred packets.





BYNETIF_COLLISION_RATEThe number of physical collisions per second on the network interfaceduring the interval. A rising rate of collisions versus outbound packets isan indication that the network is becoming increasingly congested. Thismetric does not currently include deferred packets.




101



BYNETIF_COLLISION_RATE_CUMThe average number of physical collisions per second on the networkinterface over the cumulative collection time. A rising rate of collisionsversus outbound packets is an indication that the network is becomingincreasingly congested.






BYNETIF_ERRORThe number of physical errors that occurred on the network interfaceduring the interval. An increasing number of errors may indicate ahardware problem in the network.


Logical statistics are packets seen only by the Interface Protocol (IP) layerof the networking subsystem. Not all packets seen by IP will go out andcome in through a network driver. Examples cases are the 127.0.0.1(loopback interface). Pings or other network generating commands (ftp,rlogin, and so forth) to 127.0.0.1 will not change physical driver statistics.


102

Pings to IP addresses on remote systems will change physical driverstatistics.


For HP-UX 10.20 and earlier releases, this is the same as the sum of“Ierrs” and “Oerrs” from the “netstat -i” command for a network device.See also netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Inbound Errors” and “Outbound Errors” values from the output of the“lanadmin” utility for the network interface. Remember that “lanadmin”reports cumulative counts. For this release and beyond, “netstat -i” showsnetwork activity on the logical level (IP) only.


BYNETIF_ERROR_1_MIN_RATEThe number of physical errors per minute on the network interface duringthe interval.





BYNETIF_ERROR_RATEThe number of physical errors per second on the network interface duringthe interval.



103




BYNETIF_ERROR_RATE_CUMThe average number of physical errors per second on the networkinterface over the cumulative collection time.






BYNETIF_INTERVALThe amount of time in the interval.

BYNETIF_INTERVAL_CUMThe amount of time over the cumulative collection time.


104

BYNETIF_IN_BYTE_RATEThe number of KBs per second received from the network via thisinterface during the interval. Only the bytes in packets that carry data areincluded in this rate.






BYNETIF_IN_BYTE_RATE_CUMThe average number of KBs per second received from the network via thisinterface over the cumulative collection time. Only the bytes in packetsthat carry data are included in this rate.






105



BYNETIF_IN_PACKETThe number of successful physical packets received through the networkinterface during the interval. Successful packets are those that have beenprocessed without errors or collisions.




For HP-UX 10.20 and earlier releases, this is the same as the “Ipkts”column from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Inbound Unicast Packets” and “Inbound Non-Unicast Packets” valuesfrom the output of the “lanadmin” utility for the network interface.Remember that “lanadmin” reports cumulative counts. For this releaseand beyond, “netstat -i” shows network activity on the logical level (IP)only.


BYNETIF_IN_PACKET_RATEThe number of successful physical packets per second received throughthe network interface during the interval. Successful packets are thosethat have been processed without errors or collisions.


Logical statistics are packets seen only by the Interface Protocol (IP) layerof the networking subsystem. Not all packets seen by IP will go out and


106

come in through a network driver. Examples cases are the 127.0.0.1(loopback interface). Pings or other network generating commands (ftp,rlogin, and so forth) to 127.0.0.1 will not change physical driver statistics.Pings to IP addresses on remote systems will change physical driverstatistics.



BYNETIF_IN_PACKET_RATE_CUMThe average number of physical packets per second received through thenetwork interface over the cumulative collection time.






BYNETIF_LOGL_INTERVALThe amount of time in the interval.

On systems with large numbers of Interface Protocol (IP) addresses, themeasurement code now dynamically determines the interval for updatingthe BYNETIF_LOGL_* metrics. This reduces the collection overhead forthese metrics in Glance and GPM. For the interval, it looks at how manyIP addresses there are. The update interval for the BYNETIF_LOGL_*metrics is then set as follows:

* For 1 - 20 IP addresses, the


107

counters are updated at the normal

sampling interval.

* For 21 - 120 IP addresses, the

counters are updated at an

interval (in seconds) equal to the

number of IP addresses.

* For more than 120 IP addresses,

the counters are updated every 120

seconds.

For example, if Glance or GPM is run with 5-second update intervals on an11.0 system with 200 IP addresses configured, the information shown inthe Network detail screens will only change once every 2 minutes. Thedata reflects all activity over that time so no information is lost.


BYNETIF_LOGL_INTERVAL_CUMThe amount of time over the cumulative collection time.

On HP-UX 11.0 and beyond for Glance and GPM, this metric is updatedat the BYNETIF_LOGL_INTERVAL time. On systems with large numbersof IP addresses, the BYNETIF_LOGL_INTERVAL can be greater than thesampling interval.


BYNETIF_LOGL_IN_PACKETThe number of successful logical packets received through the logicalinterface during the interval.




This is the same as the “Ipkts” column from the “netstat -i” command for anetwork device. See also netstat(1).


108



BYNETIF_LOGL_IN_PACKET_RATEThe number of successful logical packets per second received through thelogical interface during the interval.






BYNETIF_LOGL_IN_PACKET_RATE_CUMThe average number of logical packets per second received through thelogical interface over the cumulative collection time.



Logical statistics are packets seen only by the Interface Protocol (IP) layerof the networking subsystem. Not all packets seen by IP will go out andcome in through a network driver. Examples cases are the 127.0.0.1(loopback interface). Pings or other network generating commands (ftp,


109

rlogin, and so forth) to 127.0.0.1 will not change physical driver statistics.Pings to IP addresses on remote systems will change physical driverstatistics.




BYNETIF_LOGL_IP_ADDRESSThe Internet IP address of this logical network interface.


BYNETIF_LOGL_NAMEThe name of the logical network interface. These are the same names thatappear in the “Name” column of the “netstat -i” command output.


BYNETIF_LOGL_OUT_PACKETThe number of successful logical packets sent through the logical interfaceduring the interval.




This is the same as the “Opkts” column from the “netstat -i” command for anetwork device. See also netstat(1).

On HP-UX 11.0 and beyond for Glance and GPM, this metric is updatedat the BYNETIF_LOGL_INTERVAL time. On systems with large numbers


110

of IP addresses, the BYNETIF_LOGL_INTERVAL can be greater than thesampling interval.


BYNETIF_LOGL_OUT_PACKET_RATEThe number of successful logical packets per second sent through thelogical interface during the interval.






BYNETIF_LOGL_OUT_PACKET_RATE_CUMThe average number of logical packets per second sent through the logicalinterface over the cumulative collection time.



Logical statistics are packets seen only by the Interface Protocol (IP) layerof the networking subsystem. Not all packets seen by IP will go out andcome in through a network driver. Examples cases are the 127.0.0.1(loopback interface). Pings or other network generating commands (ftp,rlogin, and so forth) to 127.0.0.1 will not change physical driver statistics.


111

Pings to IP addresses on remote systems will change physical driverstatistics.




BYNETIF_NAMEThe name of the network interface.

For HP-UX 10.20 and earlier releases, these are the same names thatappear in the “Name” column of the “netstat -i” command.

For HP-UX 11.0 and beyond, these are the same names that appear in the“Description” field of the “lanadmin” command output.

BYNETIF_NET_MTUThe size of the maximum transfer unit (MTU) for this interface.


BYNETIF_NET_SPEEDThe speed of this interface. This is the bandwidth in bits/sec.


BYNETIF_NET_TYPEThe type of network device the interface communicates through.

Lan - local area network card

Loop - software loopback

interface (not tied to a

hardware device)

Loop6 - software loopback

Interface Ipv6 (not tied)

To a hardware device)

Serial - serial modem port

Vlan - virtual lan

Wan - wide area network card

Other - hardware network interface


112

type is unknown.

BYNETIF_OUT_BYTE_RATEThe number of KBs per second sent to the network via this interfaceduring the interval. Only the bytes in packets that carry data are includedin this rate.






BYNETIF_OUT_BYTE_RATE_CUMThe average number of KBs per second sent to the network via thisinterface over the cumulative collection time. Only the bytes in packetsthat carry data are included in this rate.





113




BYNETIF_OUT_PACKETThe number of successful physical packets sent through the networkinterface during the interval. Successful packets are those that have beenprocessed without errors or collisions.




For HP-UX 10.20 and earlier releases, this is the same as the “Opkts”column from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Outbound Unicast Packets” and “Outbound Non-Unicast Packets”values from the output of the “lanadmin” utility for the network interface.Remember that “lanadmin” reports cumulative counts. For this releaseand beyond, “netstat -i” shows network activity on the logical level (IP)only.


BYNETIF_OUT_PACKET_RATEThe number of successful physical packets per second sent through thenetwork interface during the interval. Successful packets are those thathave been processed without errors or collisions.

Physical statistics are packets recorded by the network drivers. Thesenumbers most likely will not be the same as the logical statistics. On HP-UX 11.0 and beyond, the values returned for the loopback interface will


114

show "na" for the physical statistics since there is no network driveractivity.




BYNETIF_OUT_PACKET_RATE_CUMThe average number of successful physical packets per second sentthrough the network interface over the cumulative collection time.






BYNETIF_PACKET_RATEThe number of successful physical packets per second sent and receivedthrough the network interface during the interval. Successful packets arethose that have been processed without errors or collisions.

Physical statistics are packets recorded by the network drivers. Thesenumbers most likely will not be the same as the logical statistics. On HP-


115

UX 11.0 and beyond, the values returned for the loopback interface willshow "na" for the physical statistics since there is no network driveractivity.




BYNETIF_QUEUEThe length of the outbound queue at the time of the last sample. Thismetric will be the same as the “Outbound Queue Length” values from theoutput of “lanadmin” utility.


This metric is only available for LAN interfaces. For WAN (Wide-AreaNetwork) interfaces such as ATM and X.25, with interface names such asel, cip/ixe, and netisdn, this metric is always equal to zero.


BYNFSOP_CLIENT_COUNTThe number of operations that the local machine processed as a client forthe current host during the interval.

BYNFSOP_CLIENT_COUNT_CUMThe number of operations that the local machine processed as a client forthe current host over the cumulative collection time.


BYNFSOP_CLIENT_TIMEThe time, in seconds, spent to service an NFS operation (as an NFSclient) during the last interval. This is measured from the time theoperation gets onto the physical network until the time a reply is received


116

from the network. In other words, this is the “service time” less the localmachine's software overhead.

BYNFSOP_CLIENT_TIME_CUMThe time, in seconds, spent to service an NFS operation (as an NFSclient) over the cumulative collection time.


This is measured from the time the operation gets onto the physicalnetwork until the time a reply is received from the network. In other words,this is the “service time” less the local machine's software overhead.

BYNFSOP_INTERVALThe amount of time in the interval.

BYNFSOP_INTERVAL_CUMThe amount of time over the cumulative collection time.


BYNFSOP_NAMEString mnemonic for the NFS operation. One of the following:

NFS Version 2

Name Operation/Action

------------------------------------

getattr Return the current

attributes of a file.

setattr Set the attributes of a

file and returns the new

attributes.

lookup Return the attributes of

a file.

readlink Return the string in the

symbolic link of a file.

read Return data from a file.

write Put data into a file.


117

create Create a file.

remove Remove a file.

rename Give a file a new name.

link Create a hard link to a

file.

symlink Create a symbolic link

to a file.

mkdir Create a directory.

rmdir Remove a directory.

readdir Read a directory entry.

statfs Return mounted file

system information.

null Verify NFS services.

No actual work done.

writecache Not used in HP-UX.

root Not used in HP-UX.

NFS Version 3 for HP-UX 10.20 and

beyond


------------------------------------





attributes.


a file.

access Check access permissions

of a user.






mkdir Make a directory.


to a file.

mknod Create a special device.





file.


readdirplus Extended read of a

directory entry.

fsstat Get dynamic file


118

system information.

fsinfo Get static file

system information.

pathconf Retrieve POSIX

information.

commit Commit cached data on

server to stable

storage.



BYNFSOP_SERVER_COUNTThe number of NFS operations that the local machine performed as aserver to the current host for this current operation type during the interval.

BYNFSOP_SERVER_COUNT_CUMThe number of NFS operations that the local machine performed as aserver to the current host for this operation type over the cumulativecollection time.


BYNFSOP_SERVER_TIMEThe time, in seconds, that the local machine spent servicing each NFSoperation as a NFS server for the current host during the interval. This ismeasured from the time the operation gets onto the physical network untilthe time a reply is received from the network. In other words, this is the“service time” less the local machine's software overhead.

BYNFSOP_SERVER_TIME_CUMThe time, in seconds, that the local machine spent servicing each NFSoperation as a NFS server for the current host over the cumulativecollection time.




119

BYNFS_CLIENT_PHYS_TIMEThe time, in seconds, that the local machine spent to service all NFSoperations (as an NFS client) to this host entry during the interval.


BYNFS_CLIENT_PHYS_TIME_CUMThe time, in seconds, that the local machine spent to service all NFSoperations (as a NFS client) to this host entry over the cumulativecollection time.



BYNFS_CLIENT_READ_BYTE_RATEThe number of KBs per second transferred during the interval by the NFSread operations where the local machine was acting as a client for thishost.

BYNFS_CLIENT_READ_BYTE_RATE_CUMThe average number of KBs per second transferred by the NFS readoperations where the local machine was acting as a client for this hostentry over the cumulative collection time.


BYNFS_CLIENT_READ_RATEThe number of NFS read operations per second where the local machinewas acting as a client to this NFS host entry during the interval.

NFS Version 2 read operations consist of getattr, lookup, readlink,readdir, null, root, statfs, and read.

NFS Version 3 read operations consist of getattr, lookup, access, readlink,read, readdir, readdirplus, fsstat, fsinfo, and null.


120

BYNFS_CLIENT_READ_RATE_CUMThe average number of NFS read operations per second where the localmachine was acting as a client to this NFS host entry over the cumulativecollection time.




BYNFS_CLIENT_SERVICEThe number of NFS IO operations processed by the local machine actingas a client for this host entry during the interval. This is sometimesreferred to as the “service count.”

BYNFS_CLIENT_SERVICE_CUMThe number of NFS IO operations processed by the local machine actingas a client to this host entry over the cumulative collection time. This issometimes referred to as the “service count.”


BYNFS_CLIENT_SERVICE_QUEUEThe local machine's number of pending NFS client read or write operationsto this NFS host at the end of the interval. This value increases as theservice time to the NFS host increases and/or as the rate of client requestsincreases.

A large value is an indication that either the NFS server is busy, or thelocal machine is a heavy user of the current server, or both.

BYNFS_CLIENT_SERVICE_QUEUE_CUMThe local machine's average number of pending NFS client read or writeoperations to this NFS host over the cumulative collection time.



121

The length of this queue increases as the service time to the NFS hostincreases and/or as the rate of the local machine's requests increases. Alarge value is an indication that either the NFS server is busy, or the localmachine is a heavy user of the current server, or both.

BYNFS_CLIENT_SERVICE_TIMEThe time, in seconds, spent for the local machine acting as an client toservice all NFS operations for this host entry during the interval.

This is the time from the point that the local machine (as a client)originates the request to the point a reply is received including IObuffering, NFS and network software layer delays, physical networklatency, and NFS server service time. This is sometimes referred to as“service time” and can be thought of as the round-trip time.

BYNFS_CLIENT_SERVICE_TIME_CUMThe time, in seconds, spent for the local machine acting as a NFS clientfor this host entry to service all NFS operations over the cumulativecollection time.


This is the time from the point that the local machine (as a client)originates the request to the point a reply is received including IObuffering, NFS and network software layer delays, physical networklatency, and NFS server service time. This is sometimes referred to as“service time” and can be thought of as the round-trip time.

BYNFS_CLIENT_WRITE_BYTE_RATEThe number of KBs per second transferred by the NFS write operationwhere the local machine was acting as a client for this host entry duringthe interval.

BYNFS_CLIENT_WRITE_BYTE_RATE_CUMThe average number of KBs per second transferred by the NFS writeoperation where the local machine was acting as a client for this host entryover the cumulative collection time.



122

BYNFS_CLIENT_WRITE_RATEThe number of NFS write operations per second where the local machinewas acting as a client to this NFS host entry during the interval.

NFS Version 2 write operations consist of setattr, write, writecache,create, remove, rename, link, symlink, mkdir, and rmdir.

NFS Version 3 write operations consist of setattr, write, create, mkdir,symlink, mknod, remove, rmdir, rename, link, pathconf, and commit.

BYNFS_CLIENT_WRITE_RATE_CUMThe average number of NFS write operations per second where the localmachine was acting as a client to this NFS host entry over the cumulativecollection time.




BYNFS_HOSTNAMEThe Internet host name of this NFS entry.

An NFS host is added if there are already NFS directories mounted orwhenever any IO activity is seen, either server or client activity. It remainslisted as long as the current midaemon program is running even if all NFSfile systems are unmounted.

A host on the network can act both as a client, or as a server at the sametime. If an NFS host acts as both client and as a server, it is only listedonce.

BYNFS_HOST_IP_ADDRESSThe Internet host IP address of this NFS entry.

BYNFS_INTERVALThe amount of time in the interval.

BYNFS_INTERVAL_CUMThe amount of time over the cumulative collection time.


123


BYNFS_LAST_PROC_IDThe PID of the last process to generate or receive NFS traffic for this NFShost.

If the host is acting as a server (that is, the local machine is the client),then the last process may be either a user application or the biod daemon.

If the host entry is acting as a client (that is, the local machine is theserver), then this process is always the nfsd daemon.

BYNFS_SERVER_READ_BYTE_RATEThe number of KBs per second transferred during the interval by the NFSread operations where the local machine was acting as a server to thishost.

BYNFS_SERVER_READ_BYTE_RATE_CUMThe average number of KBs per second transferred by the NFS readoperations where the local machine was acting as a server to this hostentry over the cumulative collection time.


BYNFS_SERVER_READ_RATEThe number of NFS read operations per second where the local machinewas acting as a server for this NFS host entry during the interval.



BYNFS_SERVER_READ_RATE_CUMThe average number of NFS read operations per second where the localmachine was acting as a server for this NFS host entry over thecumulative collection time.



124




BYNFS_SERVER_SERVICEThe number of NFS IO operations processed by the local machine actingas a server to this host entry during the interval. This is sometimesreferred to as the “service count.”

BYNFS_SERVER_SERVICE_CUMThe number of NFS IO operations processed by the local machine actingas a server to this host entry over the cumulative collection time. This issometimes referred to as the “service count.”


BYNFS_SERVER_SERVICE_TIMEThe time, in seconds, spent for the local machine acting as a NFS serverto this host entry to process the client's operations during the interval. Thisincludes all of the time from the point that the operation is received to thepoint where a reply is sent back to the client, which includes softwareoverhead and any local disk IOs.

BYNFS_SERVER_SERVICE_TIME_CUMThe time, in seconds, spent over the cumulative collection time for thelocal machine acting as a NFS server to this host entry to process theclient's operations. This includes all of the time from the point that theoperation is received to the point where a reply is sent back to the client,which includes software overhead and any local disk IOs.



125

BYNFS_SERVER_WRITE_BYTE_RATEThe number of KBs per second transferred by the NFS write operationwhere the local machine was acting as a server to this host entry duringthe interval.

BYNFS_SERVER_WRITE_BYTE_RATE_CUMThe average number of KBs per second transferred by the NFS writeoperation where the local machine was acting as a server to this host entryover the cumulative collection time.


BYNFS_SERVER_WRITE_RATEThe number of NFS write operations per second where the local machinewas acting as a server for this NFS host entry during the interval.



BYNFS_SERVER_WRITE_RATE_CUMThe average number of NFS write operations per second where the localmachine was acting as a server for this NFS host entry over thecumulative collection time.




BYOP_CLIENT_COUNTThe number of current NFS operations that the local machine hasprocessed as a NFS client during the interval.

A host on the network can act both as a client, or as a server at the sametime.



126


BYOP_CLIENT_COUNT_CUMThe number of current NFS operations that the local machine hasprocessed as a NFS client over the cumulative collection time.



BYOP_INTERVALThe amount of time in the interval.

BYOP_INTERVAL_CUMThe amount of time over the cumulative collection time.


BYOP_NAMEString mnemonic for the NFS operation. One of the following:

For NFS Version 2


------------------------------------





attributes.


a file.







127




file.


to a file.

mkdir Create a directory.



statfs Return mounted file

system information.



writecache Not used in HP-UX.

root Not used in HP-UX.

NFS Version 3 for HP-UX 10.20 and

beyond


------------------------------------





attributes.


a file.

access Check access permissions

of a user.






mkdir Make a directory.


to a file.

mknod Create a special device.





file.


readdirplus Extended read of a

directory entry.

fsstat Get dynamic file

system information.


128

fsinfo Get static file

system information.

pathconf Retrieve POSIX

information.

commit Commit cached data on

server to stable

storage.



BYOP_SERVER_COUNTThe number of current NFS operations that the local machine hasprocessed as a NFS server during the interval.


BYOP_SERVER_COUNT_CUMThe number of current NFS operations that the local machine hasprocessed as a NFS server over the cumulative collection time.



BYSWP_SWAP_PRIThe priority of this swap device. Swap space is used by the lower valuepriorities first. Since device swap is faster than file system swap, it isadvisable to have lower values for device swap.

This value is set by either the swapon(1M) command, or by the “pri=” fieldin /etc/fstab. The legal values for priority range from 0 to 10.

The “memory” swap area has a priority of 11, which is one greater than themaximum legal priority value that a real swap device can be assigned.This indicates that using memory as a swap area is only done after allother swap resources have been exhausted. This is true in extreme casesof memory pressure forcing the kernel to swap the entire process to disk.In cases of process deactivation, the memory pseudo swap actually hasthe highest priority - deactivated pages are not moved - they are simplymarked as deactivated and the space they occupy is considered pseudoswap.

BYSWP_SWAP_SPACE_AVAILThe capacity (in MB) for swapping in this swap area.


129

For “device” type swap, this value is constant. However, for “filesys” swapthis value grows as needed. File system swap grows in units of“SWCHUNKS” x DEV_BSIZE bytes, which is typically 2MB. This metric issimilar to the “AVAIL” parameters returned from /usr/sbin/swapinfo.

For “memory” type swap, this value also grows as needed or as possible,given that any memory reserved for swap cannot be used for normalvirtual memory.

Note that this is potential swap space. Since swap is allocated in fixed(SWCHUNK) sizes, not all of this space may actually be usable. Forexample, on a 61 MB disk using 2 MB swap size allocations, 1 MBremains unusable and is considered wasted space.

This metric is updated every 30 seconds or the sampling interval,whichever is greater.

BYSWP_SWAP_SPACE_NAMEThe name of the device file or file system where the swap space islocated.

Part of the system's physical memory may be allocated as a pseudo-swapdevice. It is enabled by setting the “SWAPMEM_ON” kernel parameter to1.

BYSWP_SWAP_SPACE_USEDThe amount of swap space (in MB) used in this area.

This value is similar to the “USED” column returned by the/usr/sbin/swapinfo command.


BYSWP_SWAP_TYPEThe type of swap space -- device, file system (“filesys”), remote ormemory. “Device” swap is accessed directly without going through the filesystem, and is therefore faster than “filesys” swap. “Remote” swap is usedfor diskless clients that are swapping to their diskless server. “Memory”swap is space in the system's physical memory reserved for pseudo-swapfor running processes. Using pseudo-swap means the pages are simplylocked in memory rather than copied to a swap area.

FSDETAIL_LABELThe type of entry this file system is associated with. It could be a device,partition, logical volume, or volume group.


130

FSDETAIL_NAMEThe name of the device, partition, logical volume, or volume group this filesystem is associated with.

FS_BLOCK_SIZEThe maximum block size of this file system in (KB).

A value of “na” is displayed if the file system is not mounted.

FS_DEVNAMEThe path name string of the current device.

This is the “fsname” parameter in the mount(1M) command. For NFSdevices, this includes the name of the node exporting the file system.

It is possible that a process may mount a device using the mount(2)system call. This call does not update the “/etc/mnttab” and its name isblank. This situation is rare, and should be corrected by syncer(1M).

Note that once a device is mounted, its entry is displayed, even after thedevice is unmounted, until the midaemon process terminates.

FS_DEVNOThe major and minor number of the file system.

FS_DIRNAMEThe path name of the mount point of the file system if the logical volumehas a mounted file system. This is the directory parameter of themount(1M) command for most entries.

Exceptions:

* For lvm swap areas, this field

contains “lvm swap device”.

* For logical volumes with no

mounted file systems, this field

contains “Raw Logical Volume”

(relevant only to MeasureWare

Agent).

The file names are in the same order as shown in the /usr/sbin/mount -pcommand. File systems are not displayed until they exhibit IO activityonce the midaemon has been started. Also, once a device is displayed, itcontinues to be displayed (even after the device is unmounted) until themidaemon process terminates.


131

FS_FILE_IO_RATEThe number of file system related physical IOs per second directed to thisfile system during the interval.

This value is similar to the values returned by the vmstat -d commandexcept that vmstat reports all IOs and does not break them out by filesystem. Also, vmstat reports IOs from the kernel's view, which may getbroken down by the disk driver into multiple physical IOs. Since this metricreports values from the disk driver's point of view, it is more accurate thanvmstat.

FS_FILE_IO_RATE_CUMThe average number of file IOs per second directed to this file system overthe cumulative collection time.


This value is similar to the values returned by the vmstat -d commandexcept that vmstat reports all IOs and does not break them out by filesystem. Also, vmstat reports IOs from the kernel's view, which may getbroken down by the disk driver into multiple physical IOs. Since this metricreports values from the disk driver's point of view, it is more accurate thanvmstat.

FS_FRAG_SIZEThe fragment size of this file system in (KB). A value of “na” is displayed ifthe file system is not mounted.

FS_INTERVALThe amount of time in the interval.

FS_INTERVAL_CUMThe amount of time over the cumulative collection time.


FS_IS_LVMReturns true (1) if this file system is a logical volume or 0 if a hard-partitioned file system.


132

FS_LOGL_IO_RATEThe number of logical IOs per second directed to this file system duringthe interval. Logical IOs are generated by calling the read() or write()system calls.

FS_LOGL_IO_RATE_CUMThe average number of logical IOs per second directed to this file systemover the cumulative collection time. Logical IOs are generated by callingthe read() or write() system calls.


FS_LOGL_READ_BYTE_RATEThe number of logical read KBs per second from this file system during theinterval.

FS_LOGL_READ_BYTE_RATE_CUMThe average number of logical read KBs per second from this file systemover the cumulative collection time. Logical reads are generated by callingthe read() system call.


FS_LOGL_READ_RATEThe number of logical reads per second directed to this file system duringthe interval. Logical reads are generated by calling the read() system call.

FS_LOGL_READ_RATE_CUMThe average number of logical reads per second directed to this filesystem over the cumulative collection time. Logical reads are generatedby calling the read() system call.



133

FS_LOGL_WRITE_BYTE_RATEThe number of logical writes KBs per second to this file system during theinterval.

FS_LOGL_WRITE_BYTE_RATE_CUMThe average number of logical write KBs per second to this file systemover the cumulative collection time. Logical writes are generated bycalling the write() system call.


FS_LOGL_WRITE_RATEThe number of logical writes per second directed to this file system duringthe interval. Logical writes are generated by calling the write() system call.

FS_LOGL_WRITE_RATE_CUMThe average number of logical writes per second directed to this filesystem over the cumulative collection time. Logical writes are generatedby calling the write() system call.


FS_MAX_SIZEMaximum number (of MB) that this file system could obtain if full.

This metric is updated at 4 minute intervals to minimize collectionoverhead. Note that this is the user space capacity - it is the file systemspace accessible to non root users. The bdf command shows the total filesystem capacity which includes the extra file system space accessible toroot users only.

For HP-UX 10.20 and beyond, a value of “na” may be displayed if the filesystem is not mounted.

FS_PHYS_IO_RATEThe number of physical IOs per second directed to this file system duringthe interval.


134

FS_PHYS_IO_RATE_CUMThe average number of physical IOs per second directed to this filesystem over the cumulative collection time.


FS_PHYS_READ_BYTE_RATEThe number of physical KBs per second read from this file system duringthe interval.

FS_PHYS_READ_BYTE_RATE_CUMThe average number of KBs per second of physical reads from this filesystem over the cumulative collection time.


FS_PHYS_READ_RATEThe number of physical reads per second directed to this file systemduring the interval. Physical reads are generated by user file access,virtual memory access (paging), file system management, or raw deviceaccess.

FS_PHYS_READ_RATE_CUMThe average number of physical reads per second directed to this filesystem over the cumulative collection time.


FS_PHYS_WRITE_BYTE_RATEThe number of physical KBs per second written to this file system duringthe interval.


135

FS_PHYS_WRITE_BYTE_RATE_CUMThe average number of KBs per second of physical writes to this filesystem over the cumulative collection time.


FS_PHYS_WRITE_RATEThe number of physical writes per second directed to this file systemduring the interval.

FS_PHYS_WRITE_RATE_CUMThe average number of physical writes per second directed to this filesystem over the cumulative collection time.


FS_SPACE_UTILPercentage of the file system space in use during the interval. This metricis updated at 4 minute intervals to minimize collection overhead. Note thatthis is the user space utilization - that is, it's the percentage of used filesystem space of the total user space available, not the total file systemcapacity which includes space accessible to root users only.

FS_TYPEA string indicating the file system type. Possible types are:

hfs - user file system

nfs - network file system

cdfs - CD-ROM file system

vxfs - journal file system

ffs - OSF's ffs (file on file

mount) file system

lofs - nfs loopback file system

pipefs - memory based unnamed

pipes

ufs - user file system

pc - pc file system

dcfs - data compression file


136

system

nfs3 - network file system

Version 3 for HPUX 10.20

and beyond

mvfs - multi-version file

system

FS_VM_IO_RATEThe number of virtual memory IOs per second directed to this file systemduring the interval.

FS_VM_IO_RATE_CUMThe average number of virtual memory IOs per second directed to this filesystem over the cumulative collection time.


GBL_ACTIVE_CPUThe number of CPUs online on the system.

The sar(1M) command allows you to check the status of the system CPUs.

GBL_ACTIVE_PROCAn active process is one that exists and consumes some CPU time.GBL_ACTIVE_PROC is the sum of the alive-process-time/interval-timeratios of every process that is active (uses any CPU time) during aninterval.

The following diagram of a four second interval during which twoprocesses exist on the system should be used to understand the abovedefinition. Note the difference between active processes, which consumeCPU time, and alive processes which merely exist on the system.

----------- Seconds -----------

1 2 3 4

Proc

---- ---- ---- ---- ----



Process A is alive for the entire four second interval but consumes noCPU. A's contribution to GBL_ALIVE_PROC is 4*1/4. A contributes 0*1/4


137

to GBL_ACTIVE_PROC. B's contribution to GBL_ALIVE_PROC is 3*1/4.B contributes 2*1/4 to GBL_ACTIVE_PROC. Thus, for this interval,GBL_ACTIVE_PROC equals 0.5 and GBL_ALIVE_PROC equals 1.75.

Because a process may be alive but not active, GBL_ACTIVE_PROC willalways be less than or equal to GBL_ALIVE_PROC.

This metric is a good overall indicator of the workload of the system. Anunusually large number of active processes could indicate a CPUbottleneck.

To determine if the CPU is a bottleneck, compare this metric withGBL_CPU_TOTAL_UTIL and GBL_RUN_QUEUE. IfGBL_CPU_TOTAL_UTIL is near 100 percent and GBL_RUN_QUEUE isgreater than one, there is a bottleneck.

GBL_ALIVE_PROCAn alive process is one that exists on the system. GBL_ALIVE_PROC isthe sum of the alive-process-time/interval-time ratios for every process.

The following diagram of a four second interval during which twoprocesses exist on the system should be used to understand the abovedefinition. Note the difference between active processes, which consumeCPU time, and alive processes which merely exist on the system.

----------- Seconds -----------

1 2 3 4

Proc

---- ---- ---- ---- ----



Process A is alive for the entire four second interval but consumes noCPU. A's contribution to GBL_ALIVE_PROC is 4*1/4. A contributes 0*1/4to GBL_ACTIVE_PROC. B's contribution to GBL_ALIVE_PROC is 3*1/4.B contributes 2*1/4 to GBL_ACTIVE_PROC. Thus, for this interval,GBL_ACTIVE_PROC equals 0.5 and GBL_ALIVE_PROC equals 1.75.

Because a process may be alive but not active, GBL_ACTIVE_PROC willalways be less than or equal to GBL_ALIVE_PROC.

GBL_BLANKA string of blanks.

GBL_CACHE_QUEUEThe average number of processes or kernel threads blocked on CACHE(waiting for the file system buffer cache to be updated) during the interval.Processes or kernel threads doing raw IO to a disk are not included in this


138

measurement. As this number rises, it is an indication of a disk or memorybottleneck.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on CACHE divided by the interval time.

The Global QUEUE metrics, which are based on block states, representthe average number of process or kernel thread counts, not actual queues.

The Global WAIT PCT metrics, which are also based on block states,represent the percentage of all processes or kernel threads that were aliveon the system.

No direct comparison is reasonable with the Application WAIT PCTmetrics since they represent percentages within the context of a specificapplication and cannot be summed or compared with global values easily.In addition, the sum of each Application WAIT PCT for all applications willnot equal 100% since these values will vary greatly depending on thenumber of processes or kernel threads in each application.


GBL_CACHE_WAIT_PCTThe percentage of time processes or kernel threads were blocked oncache (waiting for the file system buffer cache to be updated) during theinterval. Processes or kernel threads doing raw IO to a disk are notincluded in this measurement.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on CACHE divided by the accumulated time that allprocesses or kernel threads were alive during the interval.






139

GBL_CACHE_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on CACHE (waiting for the file system buffer cache to beupdated) during the interval. Processes or kernel threads doing raw IO toa disk are not included in this measurement.

GBL_CDFS_QUEUEThe average number of processes or kernel threads blocked on CDFS(waiting for their Compact Disk file system IO to complete) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on CDFS divided by the interval time.





GBL_CDFS_WAIT_PCTThe percentage of time processes or kernel threads were blocked onCDFS (waiting for their Compact Disk file system IO to complete) duringthe interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on CDFS divided by the accumulated time that allprocesses or kernel threads were alive during the interval.



No direct comparison is reasonable with the Application WAIT PCTmetrics since they represent percentages within the context of a specific


140

application and cannot be summed or compared with global values easily.In addition, the sum of each Application WAIT PCT for all applications willnot equal 100% since these values will vary greatly depending on thenumber of processes or kernel threads in each application.


GBL_CDFS_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on CDFS (waiting for their Compact Disk file system IO tocomplete) during the interval.

GBL_COMPLETED_PROCThe number of processes that terminated during the interval.

GBL_CPU_CSWITCH_TIMEThe time, in seconds, that the CPU spent context switching during theinterval.



GBL_CPU_CSWITCH_TIME_CUMThe time, in seconds, that the CPU spent context switching over thecumulative collection time. This includes context switches that result in theexecution of a different process and those caused by a process stopping,then resuming, with no other process running in the meantime.




141

GBL_CPU_CSWITCH_UTILThe percentage of time that the CPU spent context switching during theinterval.



GBL_CPU_CSWITCH_UTIL_CUMThe percentage of time that the CPU spent context switching over thecumulative collection time. This includes context switches that result in theexecution of a different process and those caused by a process stopping,then resuming, with no other process running in the meantime.



GBL_CPU_CSWITCH_UTIL_HIGHThe highest percentage of time during any one interval that the CPU spentcontext switching over the cumulative collection time. This includescontext switches that result in the execution of a different process andthose caused by a process stopping, then resuming, with no other processrunning in the meantime.



GBL_CPU_IDLE_TIMEThe time, in seconds, that the CPU was idle during the interval. This isthe total idle time, including waiting for I/O.


142

On a system with multiple CPUs, this metric is normalized. That is, theCPU used over all processors is divided by the number of processorsonline.

GBL_CPU_IDLE_TIME_CUMThe time, in seconds, that the CPU was idle over the cumulative collectiontime. This is the total idle time, including waiting for I/O.



GBL_CPU_IDLE_UTILThe percentage of time that the CPU was idle during the interval. This isthe total idle time, including waiting for I/O. This is the same as the sum ofthe "%idle" and "%wio" fields reported by "sar -u" command.


GBL_CPU_IDLE_UTIL_CUMThe percentage of time that the CPU was idle over the cumulativecollection time. This is the total idle time, including waiting for I/O.



GBL_CPU_IDLE_UTIL_HIGHThe highest percentage of time that the CPU was idle during any oneinterval over the cumulative collection time.



143


GBL_CPU_INTERRUPT_TIMEThe time, in seconds, that the CPU spent processing interrupts during theinterval.


GBL_CPU_INTERRUPT_TIME_CUMThe time, in seconds, that the CPU spent processing interrupts over thecumulative collection time.



GBL_CPU_INTERRUPT_UTILThe percentage of time that the CPU spent processing interrupts duringthe interval.


GBL_CPU_INTERRUPT_UTIL_CUMThe percentage of time that the CPU spent processing interrupts over thecumulative collection time.




144

GBL_CPU_INTERRUPT_UTIL_HIGHThe highest percentage of time that the CPU spent processing interruptsduring any one interval over the cumulative collection time.



GBL_CPU_NICE_TIMEThe time, in seconds, that the CPU was in user mode at a nice priorityduring the interval. The NICE metrics include positive nice value CPUtime only. Negative nice value CPU is broken out into NNICE (negativenice) metrics. Positive nice values range from 20 to 39. Negative nicevalues range from 0 to 19.


GBL_CPU_NICE_TIME_CUMThe time, in seconds, that the CPU was in user mode at a nice priorityover the cumulative collection time. The NICE metrics include positivenice value CPU time only. Negative nice value CPU is broken out intoNNICE (negative nice) metrics. Positive nice values range from 20 to 39.Negative nice values range from 0 to 19.



GBL_CPU_NICE_UTILThe percentage of time that the CPU was in user mode at a nice priorityduring the interval. The NICE metrics include positive nice value CPUtime only. Negative nice value CPU is broken out into NNICE (negative


145

nice) metrics. Positive nice values range from 20 to 39. Negative nicevalues range from 0 to 19.


GBL_CPU_NICE_UTIL_CUMThe percentage of time that the CPU was in user mode at a nice priorityover the cumulative collection time. The NICE metrics include positivenice value CPU time only. Negative nice value CPU is broken out intoNNICE (negative nice) metrics. Positive nice values range from 20 to 39.Negative nice values range from 0 to 19.



GBL_CPU_NICE_UTIL_HIGHThe highest percentage of time during any one interval that the CPU wasin user mode at a nice priority over the cumulative collection time. TheNICE metrics include positive nice value CPU time only. Negative nicevalue CPU is broken out into NNICE (negative nice) metrics. Positive nicevalues range from 20 to 39. Negative nice values range from 0 to 19.



GBL_CPU_NNICE_TIMEThe time, in seconds, that the CPU was in user mode at a nice prioritycomputed from processes with negative nice values during the interval.The NICE metrics include positive nice value CPU time only. Negativenice value CPU is broken out into NNICE (negative nice) metrics. Positivenice values range from 20 to 39. Negative nice values range from 0 to 19.

On a system with multiple CPUs, this metric is normalized. That is, theCPU used over all processors is divided by the number of processors


146

online. This represents the usage of the total processing capacityavailable.

GBL_CPU_NNICE_TIME_CUMThe time, in seconds, that the CPU was in user mode at a nice prioritycomputed from processes with negative nice values over the cumulativecollection time. The NICE metrics include positive nice value CPU timeonly. Negative nice value CPU is broken out into NNICE (negative nice)metrics. Positive nice values range from 20 to 39. Negative nice valuesrange from 0 to 19.



GBL_CPU_NNICE_UTILThe percentage of time that the CPU was in user mode at a nice prioritycomputed from processes with negative nice values during the interval.The NICE metrics include positive nice value CPU time only. Negativenice value CPU is broken out into NNICE (negative nice) metrics. Positivenice values range from 20 to 39. Negative nice values range from 0 to 19.


GBL_CPU_NNICE_UTIL_CUMThe percentage of time that the CPU was in user mode at a nice prioritycomputed from processes with negative nice values over the cumulativecollection time. The NICE metrics include positive nice value CPU timeonly. Negative nice value CPU is broken out into NNICE (negative nice)metrics. Positive nice values range from 20 to 39. Negative nice valuesrange from 0 to 19.




147

GBL_CPU_NNICE_UTIL_HIGHThe highest percentage of time during any one interval that the CPU wasin user mode at a nice priority computed from processes with negativenice values over the cumulative collection time. The NICE metrics includepositive nice value CPU time only. Negative nice value CPU is broken outinto NNICE (negative nice) metrics. Positive nice values range from 20 to39. Negative nice values range from 0 to 19.



GBL_CPU_NORMAL_TIMEThe time, in seconds, that the CPU was in user mode at normal priorityduring the interval. Normal priority user mode CPU excludes CPU used atreal-time and nice priorities.


GBL_CPU_NORMAL_TIME_CUMThe time, in seconds, that the CPU was in user mode at normal priorityover the cumulative collection time. Normal priority user mode CPUexcludes CPU used at real-time and nice priorities.



GBL_CPU_NORMAL_UTILThe percentage of time that the CPU was in user mode at normal priorityduring the interval. Normal priority user mode CPU excludes CPU used atreal-time and nice priorities.



148


GBL_CPU_NORMAL_UTIL_CUMThe percentage of time that the CPU was in user mode at normal priorityover the cumulative collection time. Normal priority user mode CPUexcludes CPU used at real-time and nice priorities.



GBL_CPU_NORMAL_UTIL_HIGHThe highest percentage of time that the CPU was in user mode at normalpriority during any one interval over the cumulative collection time. Normalpriority user mode CPU excludes CPU used at real-time and nice priorities.



GBL_CPU_QUEUEThe average number of processes or kernel threads using the CPU plusall of those processes or kernel threads blocked on PRIORITY (waiting fortheir priority to become high enough to get the CPU) during the interval.This metric is an indicator of CPU demands among the active processesor kernel threads.

To determine if the CPU is a bottleneck, compare this metric withGBL_CPU_TOTAL_UTIL. If GBL_CPU_TOTAL_UTIL is near 100 percentand GBL_CPU_QUEUE is greater than four, there is a high probability of aCPU bottleneck.

This is calculated as (the CPU time used plus the accumulated time thatall processes or kernel threads spent blocked on PRI (that is, priority))divided by the interval time.

The difference between this metric and GBL_PRI_QUEUE is that itincludes the processes or kernel threads using the CPU, if any.


149

For example, let's assume we're using a system with eight processors.We start eight CPU intensive processes that consume almost all of theCPU resources. The approximate values shown for the CPU relatedqueue metrics would be:

GBL_RUN_QUEUE = 1.0

GBL_PRI_QUEUE = 0.1

GBL_CPU_QUEUE = 1.0

Assume we start an additional eight CPU intensive processes. Theapproximate values now shown are:

GBL_RUN_QUEUE = 2.0

GBL_PRI_QUEUE = 8.0

GBL_CPU_QUEUE = 9.0

At this point, we have sixteen CPU intensive processes running on theeight processors. Keeping the definitions of the three queue metrics inmind, the run queue is 2 (that is, 16 / 8); the pri queue is 8 (only half of theprocesses can be active at any given time); and the cpu queue is 9 (half ofthe processes waiting in the cpu queue that are ready to run, plus one forthe active process).

This illustrates that the run queue is the average of the 1-minute loadaverages for all processors; the pri queue is the number of processes orkernel threads that are blocked on “PRI” (priority); and the cpu queue isthe number of processes or kernel threads in the cpu queue that are readyto run, including the processes or kernel threads using the CPU.





GBL_CPU_REALTIME_TIMEThe time, in seconds, that the CPU was in user mode at a realtime priorityduring the interval. Running at a realtime priority means that the processor kernel thread was run using the rtprio command or the rtprio system callto alter its priority. Realtime priorities range from zero to 127 and are


150

absolute priorities, meaning the realtime process with the lowest priorityruns as long as it wants to. Since this can have a huge impact on thesystem, the realtime CPU is tracked separately to make visible the effectof using realtime priorities.


GBL_CPU_REALTIME_TIME_CUMThe time, in seconds, that the CPU was in user mode at a realtime priorityover the cumulative collection time. Running at a realtime priority meansthat the process or kernel thread was run using the rtprio command or thertprio system call to alter its priority. Realtime priorities range from zero to127 and are absolute priorities, meaning the realtime process with thelowest priority runs as long as it wants to. Since this can have a hugeimpact on the system, the realtime CPU is tracked separately to makevisible the effect of using realtime priorities.



GBL_CPU_REALTIME_UTILThe percentage of time that the CPU was in user mode at a realtimepriority during the interval. Running at a realtime priority means that theprocess or kernel thread was run using the rtprio command or the rtpriosystem call to alter its priority. Realtime priorities range from zero to 127and are absolute priorities, meaning the realtime process with the lowestpriority runs as long as it wants to. Since this can have a huge impact onthe system, the realtime CPU is tracked separately to make visible theeffect of using realtime priorities.


GBL_CPU_REALTIME_UTIL_CUMThe percentage of time that the CPU was in user mode at a realtimepriority over the cumulative collection time. Running at a realtime prioritymeans that the process or kernel thread was run using the rtprio commandor the rtprio system call to alter its priority. Realtime priorities range fromzero to 127 and are absolute priorities, meaning the realtime process with


151

the lowest priority runs as long as it wants to. Since this can have a hugeimpact on the system, the realtime CPU is tracked separately to makevisible the effect of using realtime priorities.



GBL_CPU_REALTIME_UTIL_HIGHThe highest percentage of time that the CPU was in user mode at arealtime priority during any one interval over the cumulative collection time.Running at a realtime priority means that the process or kernel thread wasrun using the rtprio command or the rtprio system call to alter its priority.Realtime priorities range from zero to 127 and are absolute priorities,meaning the realtime process with the lowest priority runs as long as itwants to. Since this can have a huge impact on the system, the realtimeCPU is tracked separately to make visible the effect of using realtimepriorities.



GBL_CPU_SYSCALL_TIMEThe time, in seconds, that the CPU was in system mode (excludinginterrupt, context switch, trap, or vfault CPU) during the interval.


GBL_CPU_SYSCALL_TIME_CUMThe time, in seconds, that the CPU was in system mode (excludinginterrupt, context switch, trap, or vfault CPU) over the cumulative collectiontime.



152



GBL_CPU_SYSCALL_UTILThe percentage of time that the CPU was in system mode (excludinginterrupt, context switch, trap, or vfault CPU) during the interval.


GBL_CPU_SYSCALL_UTIL_CUMThe percentage of time that the CPU was in system mode (excludinginterrupt, context switch, trap, or vfault CPU) over the cumulative collectiontime.



GBL_CPU_SYSCALL_UTIL_HIGHThe highest percentage of time that the CPU was in system mode(excluding interrupt, context switch, trap, or vfault CPU) during any oneinterval over the cumulative collection time.




153

GBL_CPU_SYS_MODE_TIMEThe time, in seconds, that the CPU was in system mode during theinterval. A UNIX process operates in either system mode (also calledkernel mode) or user mode. When a process requests services from theoperating system with a system call, it switches into the machine'sprivileged protection mode and runs in system mode.


GBL_CPU_SYS_MODE_TIME_CUMThe time, in seconds, that the CPU was in system mode since over thecumulative collection time. A UNIX process operates in either systemmode (also called kernel mode) or user mode. When a process requestsservices from the operating system with a system call, it switches into themachine's privileged protection mode and runs in system mode.



GBL_CPU_SYS_MODE_UTILPercentage of time the CPU was in system mode during the interval. AUNIX process operates in either system mode (also called kernel mode) oruser mode. When a process requests services from the operating systemwith a system call, it switches into the machine's privileged protectionmode and runs in system mode.

This metric is a subset of the GBL_CPU_TOTAL_UTIL percentage.

This is NOT a measure of the amount of time used by system daemonprocesses, since most system daemons spend part of their time in usermode and part in system calls, like any other process.


High system mode CPU percentages are normal for IO intensiveapplications. Abnormally high system mode CPU percentages canindicate that a hardware problem is causing a high interrupt rate. It canalso indicate programs that are not calling system calls efficiently.


154

GBL_CPU_SYS_MODE_UTIL_CUMThe percentage of time that the CPU was in system mode over thecumulative collection time. A UNIX process operates in either systemmode (also called kernel mode) or user mode. When a process requestsservices from the operating system with a system call, it switches into themachine's privileged protection mode and runs in system mode.



GBL_CPU_TOTAL_TIMEThe total time, in seconds, that the CPU was not idle in the interval.

This is calculated as

GBL_CPU_TOTAL_TIME =

GBL_CPU_USER_MODE_TIME +

GBL_CPU_SYS_MODE_TIME


GBL_CPU_TOTAL_TIME_CUMThe total time that the CPU was not idle over the cumulative collectiontime.



GBL_CPU_TOTAL_UTILPercentage of time the CPU was not idle during the interval.



155

GBL_CPU_TOTAL_UTIL =

GBL_CPU_USER_MODE_UTIL +

GBL_CPU_SYS_MODE_UTIL


GBL_CPU_TOTAL_UTIL +

GBL_CPU_IDLE_UTIL = 100%

This metric varies widely on most systems, depending on the workload. Aconsistently high CPU utilization can indicate a CPU bottleneck, especiallywhen other indicators such as GBL_RUN_QUEUE andGBL_ACTIVE_PROC are also high. High CPU utilization can also occuron systems that are bottlenecked on memory, because the CPU spendsmore time paging and swapping.

GBL_CPU_TOTAL_UTIL_CUMThe percentage of total CPU time that the processor was not idle over thecumulative collection time.



GBL_CPU_TOTAL_UTIL_HIGHThe highest percentage of total CPU time during any one interval that theprocessor was not idle over the cumulative collection time.



GBL_CPU_TRAP_TIMEThe time the CPU was in trap handler code during the interval.



156


GBL_CPU_TRAP_TIME_CUMThe time, in seconds, the CPU was in trap handler code over thecumulative collection time.



GBL_CPU_TRAP_UTILThe percentage of time the CPU was executing trap handler code duringthe interval.


GBL_CPU_TRAP_UTIL_CUMThe percentage of time the CPU was in trap handler code over thecumulative collection time.



GBL_CPU_TRAP_UTIL_HIGHThe highest percentage of time during any one interval the CPU was intrap handler code over the cumulative collection time.



157


GBL_CPU_USER_MODE_TIMEThe time, in seconds, that the CPU was in user mode during the interval.User CPU is the time spent in user mode at a normal priority, at real-timepriority, and at a nice priority.


GBL_CPU_USER_MODE_TIME_CUMThe time, in seconds, that the CPU was in user mode over the cumulativecollection time. User CPU is the time spent in user mode at a normalpriority, at real-time priority, and at a nice priority.



GBL_CPU_USER_MODE_UTILThe percentage of time the CPU was in user mode during the interval.User CPU is the time spent in user mode at a normal priority, at real-timepriority, and at a nice priority.

This metric is a subset of the GBL_CPU_TOTAL_UTIL percentage.


High user mode CPU percentages are normal for computation-intensiveapplications. Low values of user CPU utilization compared to relativelyhigh values for GBL_CPU_SYS_MODE_UTIL can indicate an applicationor hardware problem.


158

GBL_CPU_USER_MODE_UTIL_CUMThe percentage of time that the CPU was in user mode over thecumulative collection time. User CPU is the time spent in user mode at anormal priority, at real-time priority, and at a nice priority.



GBL_CPU_VFAULT_TIMEThe time, in seconds, the CPU was handling page faults during theinterval.


GBL_CPU_VFAULT_TIME_CUMThe time, in seconds, the CPU was handling page faults over thecumulative collection time.



GBL_CPU_VFAULT_UTILThe percentage of time the CPU was handling page faults during theinterval.



159

GBL_CPU_VFAULT_UTIL_CUMThe percentage of time the CPU was handling page faults over thecumulative collection time.



GBL_CPU_VFAULT_UTIL_HIGHThe highest percentage of time during any one interval the CPU washandling page faults over the cumulative collection time.



GBL_CSWITCH_RATEThe average number of context switches per second during the interval.


GBL_CSWITCH_RATE_CUMThe average number of context switches per second over the cumulativecollection time.




160

GBL_CSWITCH_RATE_HIGHThe highest number of context switches per second during any intervalover the cumulative collection time.



GBL_DISK_FS_BYTEThe number of file system KBs (or MBs if specified) physically transferredto or from the disk during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.

These are bytes transferred by user file system access and do not includebytes transferred via virtual memory IOs, system IOs (inode updates), orIOs relating to raw disk access. An exception is user files accessed viathe mmap(2) call, which will not show their bytes transferred in thiscategory. They appear under virtual memory bytes transferred.

GBL_DISK_FS_BYTE_CUMThe number of file system KBs (or MBs if specified) transferred to or fromthe disk over the cumulative collection time. Only local disks are countedin this measurement. NFS devices are excluded.



GBL_DISK_FS_IOThe total of physical file system disk reads and writes during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.



161

GBL_DISK_FS_IO_CUMThe total of file system disk physical reads and writes over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.



GBL_DISK_FS_IO_PCTThe percentage of file system generated physical IOs of the total physicalIOs during the interval. Only local disks are counted in this measurement.NFS devices are excluded.


GBL_DISK_FS_IO_PCT_CUMThe percentage of file system generated physical IOs of the total physicalIOs over the cumulative collection time. Only local disks are counted inthis measurement. NFS devices are excluded.



GBL_DISK_FS_IO_RATEThe total of file system disk physical reads and writes per second duringthe interval. Only local disks are counted in this measurement. NFSdevices are excluded.


162


GBL_DISK_FS_IO_RATE_CUMThe number of file system physical disk reads and writes per second overthe cumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.



GBL_DISK_FS_READThe number of file system disk reads during the interval. Only local disksare counted in this measurement. NFS devices are excluded.

These are physical reads generated by user file system access and donot include virtual memory reads, system reads (inode access), or readsrelating to raw disk access. An exception is user files accessed via themmap(2) call, which does not show their physical reads in this category.They appear under virtual memory reads.

GBL_DISK_FS_READ_RATEThe number of file system disk reads per second during the interval. Onlylocal disks are counted in this measurement. NFS devices are excluded.


GBL_DISK_FS_WRITEThe number of file system disk writes during the interval. Only local disksare counted in this measurement. NFS devices are excluded.

These are physical writes generated by user file system access and donot include virtual memory writes, system writes (inode updates), or writesrelating to raw disk access. An exception is user files accessed via the


163

mmap(2) call, which does not show their physical writes in this category.They appear under virtual memory writes.

GBL_DISK_FS_WRITE_RATEThe number of file system disk writes per second during the interval. Onlylocal disks are counted in this measurement. NFS devices are excluded.

These are physical writes generated by user file system access and donot include virtual memory writes, system writes (inode updates), or writesrelating to raw disk access. An exception is user files accessed via themmap(2) call, which does not show their physical writes in this category.They appear under virtual memory writes.

GBL_DISK_LOGL_BYTE_RATEThe number of KBs transferred per second via disk IO calls during theinterval. Only local disks are counted in this measurement. NFS devicesare excluded.


GBL_DISK_LOGL_IOThe number of logical IOs made during the interval. Only local disks arecounted in this measurement. NFS devices are excluded.





164

GBL_DISK_LOGL_IO_CUMThe number of logical IOs made over the cumulative collection time. Onlylocal disks are counted in this measurement. NFS devices are excluded.





GBL_DISK_LOGL_IO_RATEThe number of logical IOs per second during the interval. Only local disksare counted in this measurement. NFS devices are excluded.





165

GBL_DISK_LOGL_IO_RATE_CUMThe average number of logical IOs per second over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.





GBL_DISK_LOGL_READThe number of logical reads made during the interval. Only local disks arecounted in this measurement. NFS devices are excluded.





166

GBL_DISK_LOGL_READ_BYTEThe number of KBs transferred through logical reads during the lastinterval. Only local disks are counted in this measurement. NFS devicesare excluded.


GBL_DISK_LOGL_READ_BYTE_CUMThe number of KBs transferred through logical reads over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.



GBL_DISK_LOGL_READ_BYTE_RATEThe number of KBs transferred per second via logical reads during theinterval. Only local disks are counted in this measurement. NFS devicesare excluded.


GBL_DISK_LOGL_READ_CUMThe total number of logical reads made over the cumulative collectiontime. Only local disks are counted in this measurement. NFS devices areexcluded.




167



GBL_DISK_LOGL_READ_PCTThe percentage of logical reads of the total logical IO during the interval.




GBL_DISK_LOGL_READ_PCT_CUMThe percentage of logical reads of the total logical IOs over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.



There are several reasons why logical IOs may not correspond withphysical IOs. Logical IOs may not always result in a physical disk access,since the data may already reside in memory -- either in the buffer cache,


168

or in virtual memory if the IO is to a memory mapped file. Several logicalIOs may all map to the same physical page or block. In these two cases,logical IOs are greater than physical IOs.


GBL_DISK_LOGL_READ_RATEThe average number of logical reads per second made during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.




GBL_DISK_LOGL_READ_RATE_CUMThe average number of logical reads per second over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.



There are several reasons why logical IOs may not correspond withphysical IOs. Logical IOs may not always result in a physical disk access,since the data may already reside in memory -- either in the buffer cache,or in virtual memory if the IO is to a memory mapped file. Several logical


169

IOs may all map to the same physical page or block. In these two cases,logical IOs are greater than physical IOs.


GBL_DISK_LOGL_WRITEThe number of logical writes made during the interval. Only local disks arecounted in this measurement. NFS devices are excluded.




GBL_DISK_LOGL_WRITE_BYTEThe number of KBs transferred via logical writes during the last interval.Only local disks are counted in this measurement. NFS devices areexcluded.


GBL_DISK_LOGL_WRITE_BYTE_CUMThe number of KBs transferred via logical writes over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.



170


GBL_DISK_LOGL_WRITE_BYTE_RATEThe number of KBs per second transferred via logical writes during theinterval. Only local disks are counted in this measurement. NFS devicesare excluded.


GBL_DISK_LOGL_WRITE_CUMThe total number of logical writes made over the cumulative collectiontime. Only local disks are counted in this measurement. NFS devices areexcluded.





GBL_DISK_LOGL_WRITE_PCTThe percentage of logical writes of the logical IO during the interval. Onlylocal disks are counted in this measurement. NFS devices are excluded.



171



GBL_DISK_LOGL_WRITE_PCT_CUMThe percentage of logical writes of the total logical IO over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.





GBL_DISK_LOGL_WRITE_RATEThe average number of logical writes per second made during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.



172



GBL_DISK_LOGL_WRITE_RATE_CUMThe average number of logical writes per second of the total logical IOsover the cumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.





GBL_DISK_PHYS_BYTEThe number of KBs of data transferred to and from all local disks on thesystem during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.

It is not directly related to the number of IOs, since IO requests can be ofdiffering lengths.

This metric counts all types of physical IOs, including file system, virtualmemory, and raw IO.


173

GBL_DISK_PHYS_BYTE_RATEThe average number of KBs per second at which data was transferred toand from disks during the interval. The bytes for all types physical IOs arecounted. Only local disks are counted in this measurement. NFS devicesare excluded.

This is a measure of the physical data transfer rate. It is not directlyrelated to the number of IOs, since IO requests can be of differing lengths.

This is an indicator of how much data is being transferred to and from diskdevices. Large spikes in this metric can indicate a disk bottleneck.

This includes file system, virtual memory, and raw IO.

GBL_DISK_PHYS_IOThe number of physical IOs during the interval. Only local disks arecounted in this measurement. NFS devices are excluded.

This includes all types of physical reads and writes to disk, including virtualmemory IO and raw IO.


GBL_DISK_PHYS_IO =

GBL_DISK_FS_IO +

GBL_DISK_VM_IO +

GBL_DISK_SYSTEM_IO +

GBL_DISK_RAW_IO

GBL_DISK_PHYS_IO_CUMThe total number of physical IOs over the cumulative collection time. Onlylocal disks are counted in this measurement. NFS devices are excluded.


GBL_DISK_PHYS_IO_RATEThe number of physical IOs per second during the interval. Only localdisks are counted in this measurement. NFS devices are excluded.

This includes all types of physical IOs to disk, including virtual memory IOand raw IO.


GBL_DISK_PHYS_IO_RATE =

GBL_DISK_FS_IO_RATE +

GBL_DISK_VM_IO_RATE +


174

GBL_DISK_SYSTEM_IO_RATE +

GBL_DISK_RAW_IO_RATE

GBL_DISK_PHYS_IO_RATE_CUMThe number of physical IOs per second over the cumulative collectiontime. Only local disks are counted in this measurement. NFS devices areexcluded.


GBL_DISK_PHYS_READThe number of physical reads during the interval. This includes all typesof physical reads from disk, including VM and raw reads. Only local disksare counted in this measurement. NFS devices are excluded.

There are many reasons why there is not a direct correlation between thenumber of logical IOs and physical IOs. For example, small sequentiallogical reads may be satisfied from the buffer cache, resulting in fewerphysical IOs than logical IOs. Conversely, large logical IOs or smallrandom IOs may result in more physical than logical IOs. Logical volumemappings, logical disk mirroring, and disk striping also tend to remove anycorrelation.


GBL_DISK_PHYS_READ =

GBL_DISK_FS_READ +

GBL_DISK_VM_READ +

GBL_DISK_SYSTEM_READ +

GBL_DISK_RAW_READ

GBL_DISK_PHYS_READ_BYTEThe number of KBs physically transferred from the disk during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.

GBL_DISK_PHYS_READ_BYTE_CUMThe number of KBs (or MBs if specified) physically transferred from thedisk over the cumulative collection time. Only local disks are counted inthis measurement. NFS devices are excluded.



175

GBL_DISK_PHYS_READ_BYTE_RATEThe average number of KBs transferred from the disk per second duringthe interval. Only local disks are counted in this measurement. NFSdevices are excluded.

GBL_DISK_PHYS_READ_CUMThe total number of physical reads over the cumulative collection time.Only local disks are counted in this measurement. NFS devices areexcluded.


GBL_DISK_PHYS_READ_PCTThe percentage of physical reads of total physical IO during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.

GBL_DISK_PHYS_READ_PCT_CUMThe percentage of physical reads of total physical IO over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.


GBL_DISK_PHYS_READ_RATEThe number of physical reads per second during the interval. Only localdisks are counted in this measurement. NFS devices are excluded.

This includes all types of physical reads from disk, including VM and rawreads.


GBL_DISK_PHYS_READ_RATE =

GBL_DISK_FS_READ_RATE +

GBL_DISK_VM_READ_RATE +

GBL_DISK_SYSTEM_READ_RATE +

GBL_DISK_RAW_READ_RATE


176

GBL_DISK_PHYS_READ_RATE_CUMThe average number of physical reads per second over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.


GBL_DISK_PHYS_WRITEThe number of physical writes during the interval. This includes all typesof physical writes to disk, including virtual memory and raw IO. Only localdisks are counted in this measurement. NFS devices are excluded.

Since this value is reported by the drivers, multiple physical requests thathave been collapsed to a single physical operation (due to driver IOmerging) are only counted once.

There are many reasons why there is not a direct correlation betweenlogical IOs and physical IOs. For example, small logical writes may endup entirely in the buffer cache, and later generate fewer physical IOs whenwritten to disk due to the larger IO size. Or conversely, small logical writesmay require physical prefetching of the corresponding disk blocks beforethe data is merged and posted to disk. Logical volume mappings, logicaldisk mirroring, and disk striping also tend to remove any correlation.


GBL_DISK_PHYS_WRITE =

GBL_DISK_FS_WRITE +

GBL_DISK_VM_WRITE +

GBL_DISK_SYSTEM_WRITE +

GBL_DISK_RAW_WRITE

GBL_DISK_PHYS_WRITE_BYTEThe number of KBs (or MBs if specified) physically transferred to the diskduring the interval. Only local disks are counted in this measurement.NFS devices are excluded.

GBL_DISK_PHYS_WRITE_BYTE_CUMThe number of KBs (or MBs if specified) physically transferred to the diskover the cumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.



177


GBL_DISK_PHYS_WRITE_BYTE_RATEThe average number of KBs transferred to the disk per second during theinterval. Only local disks are counted in this measurement. NFS devicesare excluded.

GBL_DISK_PHYS_WRITE_CUMThe total number of physical writes over the cumulative collection time.Only local disks are counted in this measurement. NFS devices areexcluded.



GBL_DISK_PHYS_WRITE_PCTThe percentage of physical writes of total physical IO during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.


GBL_DISK_PHYS_WRITE_PCT_CUMThe percentage of physical writes of total physical IO over the cumulativecollection time. Only local disks are counted in this measurement. NFSdevices are excluded.




178

GBL_DISK_PHYS_WRITE_RATEThe number of physical writes per second during the interval. Thisincludes all types of physical writes to disk, including virtual memory IOand raw IO. Only local disks are counted in this measurement. NFSdevices are excluded.



GBL_DISK_PHYS_WRITE_RATE =

GBL_DISK_FS_WRITE_RATE +

GBL_DISK_VM_WRITE_RATE +

GBL_DISK_SYSTEM_WRITE_RATE +

GBL_DISK_RAW_WRITE_RATE

GBL_DISK_PHYS_WRITE_RATE_CUMThe number of physical writes per second over the cumulative collectiontime. Only local disks are counted in this measurement. NFS devices areexcluded.



GBL_DISK_QUEUEThe average number of processes or kernel threads blocked on disk (in a“queue” within the disk drivers waiting for their file system disk IO tocomplete) during the interval. Processes or kernel threads doing raw IO toa disk are not included in this measurement. As this number rises, it is anindication of a disk bottleneck.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on DISK divided by the interval time.



No direct comparison is reasonable with the Application WAIT PCTmetrics since they represent percentages within the context of a specificapplication and cannot be summed or compared with global values easily.In addition, the sum of each Application WAIT PCT for all applications will


179

not equal 100% since these values will vary greatly depending on thenumber of processes or kernel threads in each application.


GBL_DISK_RAW_BYTEThe number of KBs (or MBs if specified) transferred to or from a raw diskduring the interval. Only local disks are counted in this measurement.NFS devices are excluded.

GBL_DISK_RAW_BYTE_CUMThe number of KBs (or MBs if specified) transferred to or from a raw diskover the cumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.


GBL_DISK_RAW_IOThe total number of raw reads and writes during the interval. Only localdisks are counted in this measurement. NFS devices are excluded.

GBL_DISK_RAW_IO_CUMThe total number of raw IOs over the cumulative collection time. Onlylocal disks are counted in this measurement. NFS devices are excluded.


GBL_DISK_RAW_IO_PCTThe percentage of raw IOs to total physical IOs made during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.


180

GBL_DISK_RAW_IO_PCT_CUMThe percentage of physical raw IOs to total physical IOs made over thecumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.


GBL_DISK_RAW_IO_RATEThe total number of raw reads and writes per second during the interval.

Only accesses to local disk devices are counted.

GBL_DISK_RAW_IO_RATE_CUMThe average number of raw IOs over the cumulative collection time. Onlylocal disks are counted in this measurement. NFS devices are excluded.


GBL_DISK_RAW_READThe number of raw reads during the interval. Only accesses to local diskdevices are counted.

GBL_DISK_RAW_READ_RATEThe number of raw reads per second during the interval. Only accesses tolocal disk devices are counted.

GBL_DISK_RAW_WRITEThe number of raw writes during the interval. Only accesses to local diskdevices are counted.

GBL_DISK_RAW_WRITE_RATEThe number of raw writes per second during the interval. Only accessesto local disk devices are counted.


181

GBL_DISK_REM_FS_BYTEThe number of remote file system KBs (or MBs if specified) physicallytransferred to or from the remote machine during the interval.


Remote file system IO typically occurs during client file system access ofa network file system mounted on the server. A remote file system IOdoes not necessarily imply that a physical IO occurs on the remote(server) system.

GBL_DISK_REM_FS_BYTE_CUMThe number of remote file system KBs (or MBs if specified) transferred toor from the remote machine over the cumulative collection time.




GBL_DISK_REM_FS_IOThe total of remote physical file system reads and writes during the lastinterval.




182

GBL_DISK_REM_FS_IO_CUMThe total of remote file system physical reads and writes over thecumulative collection time.




GBL_DISK_REM_FS_IO_PCTThe percentage of remote file system generated physical IOs of the totalremote physical IOs during the interval.



GBL_DISK_REM_FS_IO_PCT_CUMThe percentage of remote file system generated physical IOs of the totalremote physical IOs over the cumulative collection time.





183

GBL_DISK_REM_FS_IO_RATEThe total of remote file system physical reads and writes per secondduring the interval.



GBL_DISK_REM_FS_IO_RATE_CUMThe total of remote file system physical reads and writes per second overthe cumulative collection time.




GBL_DISK_REM_LOGL_READThe number of remote logical reads made during the interval.

Remote logical IOs include all IO requests generated on a local client to aremotely mounted file system or disk. If the logical request is satisfied onthe local client (that is, the data is in a local memory buffer), a physicalrequest is not generated. Otherwise, a physical IO request is made to theremote machine to read/write the data. Note that, in either case, a logicalIO request is made.

GBL_DISK_REM_LOGL_READ_BYTEThe number of KBs transferred via remote logical reads during the lastinterval.


184


GBL_DISK_REM_LOGL_READ_BYTE_CUMThe number of KBs transferred via remote logical reads over thecumulative collection time.



GBL_DISK_REM_LOGL_READ_CUMThe total number of remote logical reads made over the cumulativecollection time.



GBL_DISK_REM_LOGL_READ_PCTThe percentage of remote logical reads to the total remote logical IOduring the interval.



185

GBL_DISK_REM_LOGL_READ_PCT_CUMThe percentage of remote logical reads of the total remote logical IO overthe cumulative collection time.



GBL_DISK_REM_LOGL_READ_RATEThe average number of remote logical reads per second made during theinterval.


GBL_DISK_REM_LOGL_READ_RATE_CUMThe average number of remote logical reads per second made over thecumulative collection time.



GBL_DISK_REM_LOGL_WRITEThe number of remote logical writes made during the interval.

Remote logical IOs include all IO requests generated on a local client to aremotely mounted file system or disk. If the logical request is satisfied onthe local client (that is, the data is in a local memory buffer), a physicalrequest is not generated. Otherwise, a physical IO request is made to the


186

remote machine to read/write the data. Note that, in either case, a logicalIO request is made.

GBL_DISK_REM_LOGL_WRITE_BYTEThe number of KBs transferred via remote logical writes during the lastinterval.


GBL_DISK_REM_LOGL_WRITE_BYTE_CUMThe number of KBs transferred via remote logical writes over thecumulative collection time.



GBL_DISK_REM_LOGL_WRITE_CUMThe total number of remote logical writes made over the cumulativecollection time.



GBL_DISK_REM_LOGL_WRITE_PCTThe percentage of remote logical writes of the total remote logical IOduring the interval.


187


GBL_DISK_REM_LOGL_WRITE_PCT_CUMThe percentage of remote logical writes of the total remote logical IO overthe cumulative collection time.



GBL_DISK_REM_LOGL_WRITE_RATEThe average number of remote logical writes per second made during thelast interval.


GBL_DISK_REM_LOGL_WRITE_RATE_CUMThe percentage of remote logical writes of the total remote logical IO overthe cumulative collection time.




188

GBL_DISK_REM_PHYS_READThe number of remote physical reads during the interval. This includes alltypes of physical reads, including VM and raw.


GBL_DISK_REM_PHYS_READ =

GBL_DISK_REM_FS_READ +

GBL_DISK_REM_VM_READ +

GBL_DISK_REM_SYSTEM_READ +

GBL_DISK_REM_RAW_READ

If an IO cannot be satisfied in a local client machine's memory buffer, aremote physical IO request is generated. This may or may not require aphysical disk IO on the remote system. In either case, the remote IOrequest is considered a physical request on the local client machine.

GBL_DISK_REM_PHYS_READ_BYTEThe number of physical read KBs during the interval.


GBL_DISK_REM_PHYS_READ_BYTE_CUMThe number of physical read KBs (or MBs if specified) since collection wasstarted or over the cumulative collection time.



GBL_DISK_REM_PHYS_READ_CUMThe total number of remote physical reads over the cumulative collectiontime.


If an IO cannot be satisfied in a local client machine's memory buffer, aremote physical IO request is generated. This may or may not require a


189

physical disk IO on the remote system. In either case, the remote IOrequest is considered a physical request on the local client machine.

GBL_DISK_REM_PHYS_READ_PCTThe percentage of remote physical reads of total remote physical IOduring the interval.


GBL_DISK_REM_PHYS_READ_PCT_CUMThe percentage of remote physical reads of total remote physical IO overthe cumulative collection time.



GBL_DISK_REM_PHYS_READ_RATEThe number of remote physical reads per second during the interval. Thisincludes all types of physical reads, including VM and raw.


GBL_DISK_REM_PHYS_READ_RATE =

GBL_DISK_REM_FS_READ_RATE +

GBL_DISK_REM_VM_READ_RATE +

GBL_DISK_REM_SYSTEM_READ_RATE +

GBL_DISK_REM_RAW_READ_RATE


GBL_DISK_REM_PHYS_READ_RATE_CUMThe average number of remote physical reads per second over thecumulative collection time.


190



GBL_DISK_REM_PHYS_WRITEThe number of physical writes during the interval. All types of remotephysical writes are counted, including VM and raw, are counted.


GBL_DISK_REM_PHYS_WRITE =

GBL_DISK_REM_FS_WRITE +

GBL_DISK_REM_VM_WRITE +

GBL_DISK_REM_SYSTEM_WRITE +

GBL_DISK_REM_RAW_WRITE


GBL_DISK_REM_PHYS_WRITE_BYTEThe number of physical write KBs (or MBs if specified) during the interval.


GBL_DISK_REM_PHYS_WRITE_BYTE_CUMThe number of physical write KBs (or MBs if specified) over the cumulativecollection time.




191

GBL_DISK_REM_PHYS_WRITE_CUMThe total number of physical writes over the cumulative collection time.



GBL_DISK_REM_PHYS_WRITE_PCTThe percentage of physical writes of total remote physical IO during theinterval.


GBL_DISK_REM_PHYS_WRITE_PCT_CUMThe percentage of physical writes of total remote physical IO over thecumulative collection time.



GBL_DISK_REM_PHYS_WRITE_RATEThe number of remote physical writes per second during the interval. Alltypes of remote physical writes, including VM and raw, are counted.


GBL_DISK_REM_PHYS_WRITE_RATE =

GBL_DISK_REM_FS_WRITE_RATE +

GBL_DISK_REM_VM_WRITE_RATE +

GBL_DISK_REM_SYSTEM_WRITE_RATE +

GBL_DISK_REM_RAW_WRITE_RATE

If an IO cannot be satisfied in a local client machine's memory buffer, aremote physical IO request is generated. This may or may not require a


192

physical disk IO on the remote system. In either case, the remote IOrequest is considered a physical request on the local client machine.

GBL_DISK_REM_PHYS_WRITE_RATE_CUMThe number of physical writes per second over the cumulative collectiontime.



GBL_DISK_REM_RAW_BYTEThe number of remote KBs (or MBs if specified) transferred to or from araw disk during the interval. Remote raw disk IO typically occurs when aclient accesses a server disk in raw mode.

GBL_DISK_REM_RAW_BYTE_CUMThe number of remote KBs (or MBs if specified) transferred to or from araw disk over the cumulative collection time. Remote raw disk IO typicallyoccurs when a client accesses a server disk in raw mode.


GBL_DISK_REM_RAW_IOThe number of remote raw IOs during the interval. Remote raw disk IOtypically occurs when a client accesses a server disk in raw mode.

GBL_DISK_REM_RAW_IO_CUMThe total number of remote raw IOs over the cumulative collection time.Remote raw disk IO typically occurs when a client accesses a server diskin raw mode.



193

GBL_DISK_REM_RAW_IO_PCTThe percentage of remote raw IOs to total remote physical disk IOs madeduring the interval. Remote raw disk IO typically occurs when a clientaccesses a server disk in raw mode.

GBL_DISK_REM_RAW_IO_PCT_CUMThe percentage of remote raw IOs to total remote physical disk IOs madeover the cumulative collection time. Remote raw disk IO typically occurswhen a client accesses a server disk in raw mode.


GBL_DISK_REM_RAW_IO_RATEThe total number of remote raw IOs per second during the interval.Remote raw disk IO typically occurs when a client accesses a server diskin raw mode.

GBL_DISK_REM_RAW_IO_RATE_CUMThe average number of remote raw IOs over the cumulative collectiontime. Remote raw disk IO typically occurs when a client accesses a serverdisk in raw mode.


GBL_DISK_REM_SYSTEM_BYTEThe number of remote KBs (or MBs if specified) transferred by the kernelfrom or to the remote machine for file system management access orupdates during the interval.

File system management IOs are the physical accesses required to obtainor update internal information about the file system structure (inodeaccess). Accesses or updates to user data are not included in this metric.



194

GBL_DISK_REM_SYSTEM_BYTE_CUMThe number of remote KBs (or MBs if specified) transferred by the kernelto or from the remote machine for file system management access orupdates over the cumulative collection time.




GBL_DISK_REM_SYSTEM_IOThe number of remote physical IOs generated by the kernel for file systemmanagement (inode accesses or updates) during the interval.


GBL_DISK_REM_SYSTEM_IO_CUMThe number of remote physical reads and writes generated by the kernelfor file system management (inode accesses or updates) over thecumulative collection time.



GBL_DISK_REM_SYSTEM_IO_PCTThe percentage of remote physical IOs generated by the kernel for filesystem management (inode accesses or updates) to the total number ofremote physical IOs during the interval.

Remote file system IO typically occurs during client file system access ofa network file system mounted on the server. A remote file system IO


195

does not necessarily imply that a physical IO occurs on the remote(server) system.

GBL_DISK_REM_SYSTEM_IO_PCT_CUMThe percentage of remote physical IOs generated by the kernel for filesystem management (inode updates) to the total number of remotephysical disk IOs over the cumulative collection time.



GBL_DISK_REM_SYSTEM_IO_RATEThe number of remote physical IOs per second generated by the kernel forfile system management (inode accesses or updates) during the interval.


GBL_DISK_REM_SYSTEM_IO_RATE_CUMThe number of remote physical reads and writes per second generated bythe kernel over the cumulative collection time.


These are IOs for file system management (inode access or updates) anddo not include IOs to user data.


GBL_DISK_REM_VM_BYTEThe number of remote virtual memory KBs (or MBs if specified) transferredto or from the remote machine during the interval.

User file data transfers are not included in this metric unless they weredone via the mmap(2) system call.


196

Remote VM IO is typically seen on a client system that is paging in textfrom or paging out data pages to a server system. Paging in from theserver system can occur when the client is loading a program whichrequires the text pages to be fetched from the server. Paging out occurswhen client system data pages are swapped out to a remote swap deviceon the server system.

GBL_DISK_REM_VM_BYTE_CUMThe number of remote virtual memory KBs (or MBs if specified) transferredto or from the remote machine over the cumulative collection time. Theseare bytes transferred due to paging or swapping.




GBL_DISK_REM_VM_IOThe total number of remote virtual memory IOs made during the interval.These are physical IOs related to paging or swapping.



GBL_DISK_REM_VM_IO_CUMThe total number of remote virtual memory IOs over the cumulativecollection time.




197


GBL_DISK_REM_VM_IO_PCTThe percentage of remote virtual memory IO requests of total remotephysical IOs during the interval.



GBL_DISK_REM_VM_IO_PCT_CUMThe percentage of remote virtual memory IOs of the total number ofremote physical IOs over the cumulative collection time.




GBL_DISK_REM_VM_IO_RATEThe number of remote virtual memory IOs per second made during theinterval. These are physical IOs related to paging, swapping, or memorymapped file allocations.


Remote VM IO is typically seen on a client system that is paging in textfrom or paging out data pages to a server system. Paging in from theserver system can occur when the client is loading a program whichrequires the text pages to be fetched from the server. Paging out occurs


198

when client system data pages are swapped out to a remote swap deviceon the server system.

GBL_DISK_REM_VM_IO_RATE_CUMThe number of remote virtual memory IOs per second made over thecumulative collection time.




GBL_DISK_SUBSYSTEM_QUEUEThe average number of processes or kernel threads blocked on the disksubsystem (in a “queue” waiting for their file system disk IO to complete)during the interval. This is the sum of processes or kernel threads in theDISK, INODE, CACHE and CDFS wait states. Processes or kernelthreads doing raw IO to a disk are not included in this measurement. Asthis number rises, it is an indication of a disk bottleneck.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (DISK + INODE + CACHE + CDFS) divided bythe interval time.






199

GBL_DISK_SUBSYSTEM_WAIT_PCTThe percentage of time processes or kernel threads were blocked on thedisk subsystem (waiting for their file system IOs to complete) during theinterval. This is the sum of processes or kernel threads in the DISK,INODE, CACHE and CDFS wait states.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (DISK + INODE + CACHE + CDFS) divided bythe accumulated time that all processes or kernel threads were aliveduring the interval.





GBL_DISK_SYSTEM_BYTEThe number of KBs (or MBs if specified) transferred by the kernel from orto the disk for file system management access or updates during theinterval.

Only local disks are counted in this measurement. NFS devices areexcluded.


GBL_DISK_SYSTEM_BYTE_CUMThe number of KBs (or MBs if specified) transferred by the kernel to orfrom disk for file system management access or updates over thecumulative collection time.



200



GBL_DISK_SYSTEM_IOThe number of physical disk IOs generated by the kernel for file systemmanagement (inode accesses or updates) during the interval.


GBL_DISK_SYSTEM_IO_CUMThe number of physical disk IOs generated by the kernel for file systemmanagement (inode accesses or updates) over the cumulative collectiontime.



GBL_DISK_SYSTEM_IO_PCTThe percentage of physical disk IOs generated by the kernel for filesystem management (inode accesses or updates) to the total number ofphysical disk IOs during the interval.


GBL_DISK_SYSTEM_IO_PCT_CUMThe percentage of physical IOs generated by the kernel for file systemmanagement (inode updates) to the total number of physical disk IOs overthe cumulative collection time.




201

GBL_DISK_SYSTEM_IO_RATEThe number of physical disk IOs per second generated by the kernel forfile system management (inode accesses or updates) during the interval.


GBL_DISK_SYSTEM_IO_RATE_CUMThe number of physical disk IOs per second generated by the kernel forfile system management (inode accesses or updates) over the cumulativecollection time. This rate does not include IOs to user data.



GBL_DISK_SYSTEM_READNumber of physical disk reads generated by the kernel for file systemmanagement (inode accesses) during the interval.


GBL_DISK_SYSTEM_READ_RATENumber of physical disk reads per second generated by the kernel for filesystem management (inode accesses) during the interval.


GBL_DISK_SYSTEM_WRITENumber of physical disk writes generated by the kernel for file systemmanagement (inode updates) during the interval.


GBL_DISK_SYSTEM_WRITE_RATENumber of physical disk writes per second generated by the kernel for filesystem management (inode updates) during the interval.


202


GBL_DISK_TIME_PEAKThe time, in seconds, during the interval that the busiest disk wasperforming IO transfers. This is for the busiest disk only, not all diskdevices. This counter is based on an end-to-end measurement for eachIO transfer updated at queue entry and exit points.


GBL_DISK_UTILThe average percentage of time during the interval that all disks had IO inprogress from the point of view of the Operating System. This is theaverage utilization for all disks. Only local disks are counted in thismeasurement. NFS devices are excluded.

GBL_DISK_UTIL_PEAKThe utilization of the busiest disk during the interval. This utilization is thepercentage of time during the interval that the busiest disk device had IOin progress from the point of view of the Operating System. It is not anaverage utilization over all the disk devices. Only local disks are countedin this measurement. NFS devices are excluded.

A peak disk utilization of more than 50 percent often indicates a disk IOsubsystem bottleneck situation. A bottleneck may not be in the physicaldisk drive itself, but elsewhere in the IO path.

GBL_DISK_UTIL_PEAK_CUMThe average utilization of the busiest disk in each interval over thecumulative collection time. Utilization is the percentage of time in useversus the time in the measurement interval. For each interval a differentdisk may be the busiest. Only local disks are counted in thismeasurement. NFS devices are excluded.


GBL_DISK_UTIL_PEAK_HIGHThe highest utilization of any disk during any interval over the cumulativecollection time. Utilization is the percentage of time in use versus the timein the measurement interval. Only local disks are counted in thismeasurement. NFS devices are excluded.


203


GBL_DISK_UTIL_PEAK_OTHERSThe non-VM IO percent of the total utilization percent of the busiest diskduring the interval. Utilization is the percentage of time in use versus thetime in the measurement interval. Only local disks are counted in thismeasurement. NFS devices are excluded.

GBL_DISK_UTIL_PEAK_VMThe VM IO percent of the total utilization percent of the busiest disk duringthe interval. Utilization is the percentage of time in use versus the time inthe measurement interval. Only local disks are counted in thismeasurement. NFS devices are excluded.

GBL_DISK_VM_BYTEThe number of virtual memory KBs (or MBs if specified) transferred to orfrom the disk during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.


GBL_DISK_VM_BYTE_CUMThe number of virtual memory KBs (or MBs if specified) transferred to orfrom the disk over the cumulative collection time. Only local disks arecounted in this measurement. NFS devices are excluded.



GBL_DISK_VM_IOThe total number of virtual memory IOs made during the interval. Onlylocal disks are counted in this measurement. NFS devices are excluded.



204

GBL_DISK_VM_IO_CUMThe total number of virtual memory IOs over the cumulative collectiontime.




GBL_DISK_VM_IO_PCTThe percentage of virtual memory IO requests of total physical disk IOsduring the interval.



GBL_DISK_VM_IO_PCT_CUMThe percentage of virtual memory IOs of the total number of physical diskIOs over the cumulative collection time.




GBL_DISK_VM_IO_RATEThe number of virtual memory IOs per second made during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.



205

GBL_DISK_VM_IO_RATE_CUMThe number of virtual memory IOs per second made over the cumulativecollection time.




GBL_DISK_VM_READThe number of virtual memory reads made during the interval. Only localdisks are counted in this measurement. NFS devices are excluded.

Reads to user file data are not included in this metric unless they wereaccessed via the mmap(2) system call.

GBL_DISK_VM_READ_CUMThe number of virtual memory reads made over the cumulative collectiontime. Only local disks are counted in this measurement. NFS devices areexcluded.



GBL_DISK_VM_READ_RATEThe number of virtual memory reads per second made during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.


GBL_DISK_VM_READ_RATE_CUMThe average number of virtual memory reads per second made over thecumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.



206



GBL_DISK_VM_READ_RATE_HIGHThe highest number of virtual memory reads per second made during anyinterval over the cumulative collection time. Only local disks are countedin this measurement. NFS devices are excluded.



GBL_DISK_VM_WRITEThe number of virtual memory writes made during the interval. Only localdisks are counted in this measurement. NFS devices are excluded.

Writes to user file data are not included in this metric unless they weredone via the mmap(2) system call.


GBL_DISK_VM_WRITE_CUMThe number of virtual memory writes made over the cumulative collectiontime. Only local disks are counted in this measurement. NFS devices areexcluded.





207

GBL_DISK_VM_WRITE_RATEThe number of virtual memory writes per second made during the interval.Only local disks are counted in this measurement. NFS devices areexcluded.



GBL_DISK_VM_WRITE_RATE_CUMThe average number of virtual memory writes per second made over thecumulative collection time. Only local disks are counted in thismeasurement. NFS devices are excluded.




GBL_DISK_VM_WRITE_RATE_HIGHThe highest number of virtual memory writes per second made during anyinterval over the cumulative collection time. Only local disks are countedin this measurement. NFS devices are excluded.




GBL_DISK_WAIT_PCTThe percentage of time processes or kernel threads were blocked on DISK(waiting in a disk driver for their disk IO to complete) during the interval.


208

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on DISK divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





GBL_DISK_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on DISK (waiting in a disk driver for their disk IO tocomplete) during the interval.

GBL_FS_SPACE_UTIL_PEAKThe percentage of occupied disk space to total disk space for the fullestfile system found during the interval. Only locally mounted file systems arecounted in this metric. CDROM and PC file systems are also excluded.

This metric can be used as an indicator that at least one file system on thesystem is running out of disk space.

This metric can exceed 100 percent. This is because a portion of the filesystem space is reserved as a buffer and can only be used by root. If theroot user has made the file system grow beyond the reserved buffer, theutilization will be greater than 100 percent. This is a dangerous situationsince if the root user totally fills the file system, the system may crash.

GBL_GRAPHICS_QUEUEThe average number of processes or kernel threads blocked on graphics(waiting for their graphics operations to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on GRAPH (that is, graphics) divide by the intervaltime.


209





GBL_GRAPHICS_WAIT_PCTThe percentage of time processes or kernel threads were blocked ongraphics (waiting for their graphics operations to complete) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on GRAPH (that is, graphics) divide by theaccumulated time that all processes or kernel threads were alive duringthe interval.






210

GBL_GRAPHICS_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on graphics (waiting for their graphics operations tocomplete) during the interval.

GBL_INODE_QUEUEThe average number of processes or kernel threads blocked on INODE(waiting for an inode to be updated or to become available) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on INODE divided by the interval time.

Inodes are used to store information about files within the file system.Every file has at least two inodes associated with it (one for the directoryand one for the file itself). The information stored in an inode includes theowners, timestamps, size, and an array of indices used to translate logicalblock numbers to physical sector numbers. There is a separate inodemaintained for every view of a file, so if two processes have the same fileopen, they both use the same directory inode, but separate inodes for thefile.





GBL_INODE_WAIT_PCTThe percentage of time processes or kernel threads were blocked onINODE (waiting for an inode to be updated or to become available) duringthe interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on INODE divided by the accumulated time that allprocesses or kernel threads were alive during the interval.

Inodes are used to store information about files within the file system.Every file has at least two inodes associated with it (one for the directory


211

and one for the file itself). The information stored in an inode includes theowners, timestamps, size, and an array of indices used to translate logicalblock numbers to physical sector numbers. There is a separate inodemaintained for every view of a file, so if two processes have the same fileopen, they both use the same directory inode, but separate inodes for thefile.





GBL_INODE_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on INODE (waiting for an inode to be updated or to becomeavailable) during the interval.

GBL_INTERRUPT_RATEThe average number of IO interrupts per second during the interval.

GBL_INTERRUPT_RATE_CUMThe average number of IO interrupts per second over the cumulativecollection time.


GBL_INTERRUPT_RATE_HIGHThe highest number of IO interrupts per second during any one intervalover the cumulative collection time.


212


GBL_INTERVALThe amount of time in the interval.

This measured interval is slightly larger than the desired or configuredinterval if the collection program is delayed by a higher priority processand cannot sample the data immediately.

GBL_INTERVAL_CUMThe amount of time over the cumulative collection time.


GBL_IPC_QUEUEThe average number of processes or kernel threads blocked onInterProcess Communication (IPC) (waiting for their interprocesscommunication calls to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on IPC divided by the interval time.






213

GBL_IPC_SUBSYSTEM_QUEUEThe average number of processes or kernel threads blocked on theInterProcess Communication (IPC) subsystems (waiting for theirinterprocess communication activity to complete) during the interval. Thisis the sum of processes or kernel threads in the IPC, MSG, SEM, PIPE,SOCKT (that is, sockets) and STRMS (that is, streams IO) wait states.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (IPC + MSG + SEM + PIPE + SOCKT +STRMS) divided by the interval time.





GBL_IPC_SUBSYSTEM_WAIT_PCTThe percentage of time processes or kernel threads were blocked on theInterProcess Communication (IPC) subsystems (waiting for theirinterprocess communication activity to complete) during the interval. Thisis the sum of processes or kernel threads in the IPC, MSG, SEM, PIPE,SOCKT (that is, sockets) and STRMS (that is, streams IO) wait states.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (IPC + MSG + SEM + PIPE + SOCKT +STRMS) divided by the accumulated time that all processes or kernelthreads were alive during the interval.





214


GBL_IPC_WAIT_PCTThe percentage of time processes or kernel threads were blocked onInterProcess Communication (IPC) (waiting for their interprocesscommunication calls to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on IPC divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





GBL_IPC_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on InterProcess Communication (IPC) (waiting for theirinterprocess communication calls to complete) during the interval.

GBL_JOBCTL_QUEUEThe average number of processes or kernel threads blocked on job control(having been stopped with the job control facilities) during the interval. Jobcontrol waits include waiting at a debug breakpoint, as well as beingblocked attempting to write (from background) to a terminal which has the“stty tostop” option set.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on job control divided by the interval time.


215





GBL_JOBCTL_WAIT_PCTThe percentage of time processes or kernel threads were blocked on jobcontrol (having been stopped with the job control facilities) during theinterval. Job control waits include waiting at a debug breakpoint, as wellas being blocked attempting to write (from background) to a terminal whichhas the “stty tostop” option set.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on job control divided by the accumulated time thatall processes or kernel threads were alive during the interval.






216

GBL_JOBCTL_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on job control (having been stopped with the job controlfacilities) during the interval. Job control waits include waiting at a debugbreakpoint, as well as being blocked attempting to write (from background)to a terminal which has the “stty tostop” option set.

GBL_LAN_QUEUEThe average number of processes or kernel threads blocked on LAN(waiting for their IO over the LAN to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on LAN divided by the interval time.





GBL_LAN_WAIT_PCTThe percentage of time processes or kernel threads were blocked on LAN(waiting for their IO over the LAN to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on LAN divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





217



GBL_LAN_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on LAN (waiting for their IO over the LAN to complete)during the interval.

GBL_LOST_MI_TRACE_BUFFERSThe number of trace buffers lost by the measurement processing daemon.If this value is > 0, the measurement subsystem is not keeping up with thesystem events that generate traces.

GBL_MACHINEA text string representing the type of computer. This is similar to what isreturned by the command “uname -m”.

GBL_MACHINE_MODELThe CPU model. This is similar to the information returned by theGBL_MACHINE metric and the “uname -m” command. However, thismetric returns more information on some processors like the T500 series.

GBL_MEM_ACTIVE_VIRTThe total virtual memory (in KBs unless otherwise specified) allocated forprocesses that are currently on the run queue or processes that haveexecuted recently. This is the sum of the virtual memory sizes of the dataand stack regions for these processes.

GBL_MEM_ACTIVE_VIRT_UTILThe percentage of total virtual memory active at the end of the interval.

Active virtual memory is the virtual memory associated with processes thatare currently on the run queue or processes that have executed recently.This is the sum of the virtual memory sizes of the data and stack regionsfor these processes.


218

GBL_MEM_AVAILThe amount of physical available memory in the system (in KBs unlessotherwise specified).

GBL_MEM_CACHEThe amount of physical memory (in KBs unless otherwise specified) usedby the buffer cache during the interval.

The buffer cache is a memory pool used by the system to stage disk IOdata for the driver.

GBL_MEM_CACHE_HITThe number of buffer cache reads resolved from the buffer cache (ratherthan going to disk) during the interval.

Buffer cache reads can occur as a result of a logical read (for example, fileread system call), a read generated by a client, a read-ahead on behalf ofa logical read or a system procedure.

This metric is obtained by measuring the number of buffered read callsthat were satisfied by the data that was in the file system buffer cache.Reads that are not in the buffer cache result in disk IO. raw IO and virtualmemory IO, are not counted in this metric.

GBL_MEM_CACHE_HIT_CUMThe number of buffer cache reads resolved from the buffer cache (ratherthan going to disk) over the cumulative collection time.


This metric is obtained by measuring the number of buffered read callsthat were satisfied by the data that was in the file system buffer cache.Reads that are not in the buffer cache result in disk IO. raw IO and virtualmemory IO, are not counted in this metric.


GBL_MEM_CACHE_HIT_PCTThe percentage of buffer cache reads satisfied in the file system buffercache (rather than going to disk) during the interval.



219

This metric is obtained by measuring the number of buffered read callsthat were satisfied by the data that was in the file system buffer cache.Reads to filesystem file buffers that are not in the buffer cache result indisk IO. Reads to raw IO and virtual memory IO (including memorymapped files), do not go through the filesystem buffer cache, and so arenot relevant to this metric.

A low cache hit rate may indicate low efficiency of the buffer cache, eitherbecause applications have poor data locality or because the buffer cacheis too small.

Overly large buffer cache sizes can lead to a memory bottleneck. Thebuffer cache should be sized small enough so that pageouts do not occureven when the system is busy.

However, in the case of VxFS, all memory-mapped IOs show up as pageins/page outs and are not a result of memory pressure.

GBL_MEM_CACHE_HIT_PCT_CUMThe average percentage of buffer cache reads resolved from the buffercache (rather than going to disk) over the cumulative collection time.



GBL_MEM_CACHE_HIT_PCT_HIGHThe highest interval percentage of buffer cache reads resolved from thebuffer cache (rather than going to disk) over the cumulative collection time.



GBL_MEM_CACHE_UTILThe percentage of physical memory used by the buffer cache during theinterval.

The buffer cache is a memory pool used by the system to stage disk IOdata for the driver.


220

GBL_MEM_CACHE_WRITE_HITThe number of write cache hits - logical writes that did not result inphysical IOs during the interval.

A cache write hit occurs when a logical write request is issued to a diskfile block that is already mapped in a buffer that is in a delayed write state.This metric gives an indication of how many physical IOs are eliminated asa result of buffering logical write requests. Physical IOs are eliminated inenvironments where asynchronous writes are done (see the O_SYNC flagin open(2)) to the same file blocks before being explicitly written to the diskor flushed to disk by the syncher process. Environments that attempt tominimize the chance of file system data loss by issuing synchronous writesor by using shorter syncer intervals will see fewer cache write hits.

During a short interval, the number of physical writes can exceed thenumber of logical write requests. This would yield a negative number of“write hits”. If this occurs in an interval, “na” will be returned.

GBL_MEM_CACHE_WRITE_HIT_CUMThe number of write cache hits - logical writes that did not result inphysical IOs over the cumulative collection time.



GBL_MEM_CACHE_WRITE_HIT_PCTThe percentage of logical disk writes that did not result in physical disk IOsduring the interval.



221

During a short interval, the number of physical writes can exceed thenumber of logical write requests. This would yield a negative number of“write hits”. If this occurs in an interval, “na” will be returned.

GBL_MEM_CACHE_WRITE_HIT_PCT_CUMThe percentage of logical disk writes that did not result in physical disk IOsover the cumulative collection time.



GBL_MEM_DNLC_HITThe number of times a pathname component was found in the directoryname lookup cache (rather than requiring a disk read to find a file) duringthe interval.

The directory name lookup cache is used to minimize sequential searchesthrough directory entries for pathname components during pathname toinode translations. Such translations are done whenever a file is accessedthrough its filename. The cache holds the inode cache table offset forrecently referenced pathname components. Pathname components thatexceed 15 characters are not cached.

Any system call that includes a path parameter can result in directoryname lookup cache activity, including but not limited to system calls suchas open, stat, exec, lstat, unlink. Each component of a path parameter isparsed and converted to an inode separately, therefore several dnlc hitsper path are possible.

High directory name cache hit rates will be seen on systems wherepathname component requests are frequently repeated. For example,when users or applications work in the same directory where theyrepeatedly list or open the same files, cache hit rates will be high.

Unusually low cache hit rates might be seen on systems where users orapplications access many different directories in no particular pattern. Lowcache hit rates can also be an indicator of an underconfigured inodecache. When an inode cache is too small, the kernel will more frequentlyhave to flush older inode cache and their corresponding directory namecache entries in order to make room for new inode cache entries.


222

The directory name lookup cache is static in size and is allocated in kernelmemory. As a result, it is not affected by user memory constraints. Thesize of the cache is stored in the kernel variable “ncsize” and is not directlytunable by the system administrator; however, it can be changed indirectlyby tuning other tables used in the formula to compute the “ncsize”. Theformula is:

ncsize = MAX(((nproc+16+maxusers)+

32+(2*npty)),ninode)

Note that ncsize is always >= ninode which is the default size of the inodecache. This is because the directory name cache contains inode tableoffsets for each cached pathname component.

GBL_MEM_DNLC_HIT_CUMThe number of times a pathname component was found in the directoryname lookup cache (rather than requiring a disk read to find a file) over thecumulative collection time.








223




GBL_MEM_DNLC_HIT_PCTThe percentage of time a pathname component was found in the directoryname lookup cache (rather than requiring a disk read to find a file) duringthe interval.










224

GBL_MEM_DNLC_HIT_PCT_CUMThe percentage of time a pathname component was found in the directoryname lookup cache (rather than requiring a disk read to find a file) over thecumulative collection time.










GBL_MEM_DNLC_HIT_PCT_HIGHThe highest percentage of time during any one interval that a pathnamecomponent was found in the directory name lookup cache (rather thanrequiring a disk read to find a file) over the cumulative collection time.



225










GBL_MEM_DNLC_LONGSThe number of times a pathname component was too long to be found inthe directory name lookup cache during the interval.


Any system call that includes a path parameter can result in directoryname lookup cache activity, including but not limited to system calls suchas open, stat, exec, lstat, unlink. Each component of a path parameter is


226

parsed and converted to an inode separately, therefore several dnlc hitsper path are possible.







GBL_MEM_DNLC_LONGS_CUMThe number of times a pathname component was too long to be found inthe directory name lookup cache over the cumulative collection time.






227






GBL_MEM_DNLC_LONGS_PCTThe percentage of time a pathname component was too long to be foundin the directory name lookup cache during the interval.







228




GBL_MEM_DNLC_LONGS_PCT_CUMThe percentage of time a pathname component was too long to be foundin the directory name lookup cache over the cumulative collection time.











229

GBL_MEM_DNLC_LONGS_PCT_HIGHThe highest percentage of time during any one interval that a pathnamecomponent was too long to be found in the directory name lookup cacheover the cumulative collection time.










GBL_MEM_FREEThe amount of memory not allocated (in KBs unless otherwise specified).As this value drops, the likelihood increases that swapping or paging out todisk may occur to satisfy new memory requests.


230

GBL_MEM_FREE_UTILThe percentage of physical memory that was free at the end of theinterval.

GBL_MEM_PAGEINThe total number of page ins from the disk during the interval. Thisincludes pages paged in from paging space and from the file system.


GBL_MEM_PAGEIN_BYTEThe number of KBs (or MBs if specified) of page ins during the interval.

GBL_MEM_PAGEIN_BYTE_CUMThe number of KBs (or MBs if specified) of page ins over the cumulativecollection time.


GBL_MEM_PAGEIN_BYTE_RATEThe number of KBs per second of page ins during the interval.

GBL_MEM_PAGEIN_BYTE_RATE_CUMThe average number of KBs per second of page ins over the cumulativecollection time.


GBL_MEM_PAGEIN_BYTE_RATE_HIGHThe highest number of KBs per second of page ins during any intervalover the cumulative collection time.



231


GBL_MEM_PAGEIN_CUMThe total number of page ins from the disk over the cumulative collectiontime.



GBL_MEM_PAGEIN_RATEThe total number of page ins per second during the interval. This includespages paged in from paging space and from the file system.

This is the same as the “page ins” value from the “vmstat -s” command.Remember that “vmstat -s” reports cumulative counts.

To determine the rate (that is, GBL_MEM_PAGEIN_RATE) for the currentinterval, subtract the previous value from the current value and then divideby the length of the interval.

Keep in mind that whenever any comparisons are made with other tools,both tools must be interval synchronized with each other in order to bevalid.

GBL_MEM_PAGEIN_RATE_CUMThe average number of page ins per second over the cumulative collectiontime. This includes pages paged in from paging space and from the filesystem.



GBL_MEM_PAGEIN_RATE_HIGHThe highest number of page ins per second from disk during any intervalover the cumulative collection time.



232

GBL_MEM_PAGEOUTThe total number of page outs to the disk during the interval. This includespages paged out to paging space and to the file system.

On HP-UX 11i, the value shown is forced page outs initiated by vhandthat are due to memory pressure. For HP-UX 11.0, the page out activitymay include memory mapped IOs on some file systems (for example,VxFS).


GBL_MEM_PAGEOUT_BYTEThe number of KBs (or MBs if specified) of page outs during the interval.



GBL_MEM_PAGEOUT_BYTE_CUMThe number of KBs (or MBs if specified) of page outs over the cumulativecollection time.



GBL_MEM_PAGEOUT_BYTE_RATEThe number of KBs (or MBs if specified) per second of page outs duringthe interval.


GBL_MEM_PAGEOUT_BYTE_RATE_CUMThe average number of KBs per second of page outs over the cumulativecollection time.



233



GBL_MEM_PAGEOUT_BYTE_RATE_HIGHThe highest number of KBs per second of page outs during any intervalover the cumulative collection time.



GBL_MEM_PAGEOUT_CUMThe total number of page outs to the disk over the cumulative collectiontime.




GBL_MEM_PAGEOUT_RATEThe total number of page outs to the disk per second during the interval.This includes pages paged out to paging space and to the file system.


This is the same as the “page outs” value from the “vmstat -s” command.Remember that “vmstat -s” reports cumulative counts.

To determine the rate (that is, GBL_MEM_PAGEOUT_RATE) for thecurrent interval, subtract the previous value from the current value andthen divide by the length of the interval.


234


GBL_MEM_PAGEOUT_RATE_CUMThe average number of page outs to the disk per second over thecumulative collection time. This includes pages paged out to pagingspace and to the file system.




GBL_MEM_PAGEOUT_RATE_HIGHThe highest number of page outs per second to disk during any intervalover the cumulative collection time.




GBL_MEM_PAGE_FAULTThe number of page faults that occurred during the interval.

GBL_MEM_PAGE_FAULT_CUMThe number of page faults that occurred over the cumulative collectiontime.



235

GBL_MEM_PAGE_FAULT_RATEThe number of page faults per second during the interval.

GBL_MEM_PAGE_FAULT_RATE_CUMThe average number of page faults per second over the cumulativecollection time.


GBL_MEM_PAGE_FAULT_RATE_HIGHThe highest page fault per second during any interval over the cumulativecollection time.


GBL_MEM_PAGE_REQUESTThe number of page requests to or from the disk during the interval.

This is the same as the sum of the “page ins” and “page outs” values fromthe “vmstat -s” command. Remember that “vmstat -s” reports cumulativecounts.

To determine the count (that is, GBL_MEM_PAGE_REQUEST) for thecurrent interval, subtract the previous value from the current value.

To determine the rate (that is, GBL_MEM_PAGE_REQUEST_RATE) forthe current interval, subtract the previous value from the current value andthen divide by the length of the interval.


GBL_MEM_PAGE_REQUEST_CUMThe total number of page requests to or from the disk over the cumulativecollection time.



236





GBL_MEM_PAGE_REQUEST_RATEThe number of page requests to or from the disk per second during theinterval.




Keep in mind that whenever any comparisons are made with other tools,both tools must be interval synchronized with each other in order to bevalid. Higher than normal rates can indicate either a memory or a diskbottleneck. Compare GBL_DISK_UTIL_PEAK and GBL_MEM_UTIL todetermine which resource is more constrained. High rates may alsoindicate memory thrashing caused by a particular application or set ofapplications. Look for processes with high major fault rates to identify theculprits.

GBL_MEM_PAGE_REQUEST_RATE_CUMThe average number of page requests to or from the disk per second overthe cumulative collection time.





237



GBL_MEM_PAGE_REQUEST_RATE_HIGHThe highest number of page requests per second during any interval overthe cumulative collection time.






GBL_MEM_PAGE_SIZE_MAXThe maximum page size allowed for a memory region on the system.


GBL_MEM_PHYSThe amount of physical memory in the system (in KBs unless otherwisespecified). Banks with bad memory are not counted.

Note that on some machines, the Processor Dependent Code (PDC) codeuses the upper 1MB of memory and thus reports less than the actualphysical memory of the system. Thus, on a system with 256MB ofphysical memory, this metric and dmesg(1M) might only report267,386,880 bytes (255MB). This is all the physical memory that softwareon the machine can access.


238

GBL_MEM_QUEUEThe average number of processes or kernel threads blocked on memory(waiting for virtual memory disk accesses to complete) during the interval.This typically happens when processes or kernel threads are allocating alarge amount of memory. It can also happen when processes or kernelthreads access memory that has been paged out to disk (swap) becauseof overall memory pressure on the system. Note that large programs canblock on VM disk access when they are initializing, bringing their text anddata pages into memory. When this metric rises, it can be an indication ofa memory bottleneck, especially if overall system memory utilization(GBL_MEM_UTIL) is near 100% and there is also swapout or page outactivity.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on memory divided by the interval time.





GBL_MEM_SWAPThe total number of deactivations/reactivations during the interval.

Process swapping has been replaced in HPUX 10.0 by a combination ofpaging and deactivation. Process deactivation occurs when the system isthrashing or when the amount of free memory falls below a critical level.The swapper then marks certain processes for deactivation and removesthem from the run queue. Pages within the associated memory regionsare reused or paged out by the memory management vhand process infavor of pages belonging to processes that are not deactivated. Unliketraditional process swapping, deactivated memory pages may or may notbe written out to the swap area, because a process could be reactivatedbefore the paging occurs.

To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swapareas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smaller


239

than HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process andreading in any pages that were actually paged out to the swap area whilethe process was deactivated.

This is the same as the sum of the “swap ins” and “swap outs” valuesfrom the “vmstat -s” command. Remember that “vmstat -s” reportscumulative counts.

This metric can be compared to the sum of the “si” and “so” values fromthe “vmstat -S” command. The “si” value reports the number of processesswapped in (or reactivated), while the “so” value reports the number ofprocesses swapped out (or deactivated) in HP-UX 10.0.

To determine the count (that is, GBL_MEM_SWAP) for the currentinterval, subtract the previous value from the current value.

To determine the rate (that is, GBL_MEM_SWAP_RATE) for the currentinterval, subtract the previous value from the current value and then divideby the length of the interval.


GBL_MEM_SWAPINThe number of reactivations during the interval.


To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swapareas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smallerthan HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process andreading in any pages that were actually paged out to the swap area whilethe process was deactivated.

This is the same as the “swap ins” value from the “vmstat -s” command.Remember that “vmstat -s” reports cumulative counts.

This metric can be compared to the “si” value from the “vmstat -S”command. The “si” value reports the number of processes swapped in (orreactivated) in HP-UX 10.0.


240

To determine the count (that is, GBL_MEM_SWAPIN) for the currentinterval, subtract the previous value from the current value.

To determine the rate (that is, GBL_MEM_SWAPIN_RATE) for the currentinterval, subtract the previous value from the current value and then divideby the length of the interval.


GBL_MEM_SWAPIN_BYTEThe number of KBs transferred in from disk due to reactivations during theinterval.



GBL_MEM_SWAPIN_BYTE_CUMThe number of KBs transferred in from disk due to reactivations over thecumulative collection time.


Process swapping has been replaced in HPUX 10.0 by a combination ofpaging and deactivation. Process deactivation occurs when the system isthrashing or when the amount of free memory falls below a critical level.The swapper then marks certain processes for deactivation and removesthem from the run queue. Pages within the associated memory regionsare reused or paged out by the memory management vhand process infavor of pages belonging to processes that are not deactivated. Unliketraditional process swapping, deactivated memory pages may or may not


241

be written out to the swap area, because a process could be reactivatedbefore the paging occurs.


GBL_MEM_SWAPIN_BYTE_RATEThe number of KBs per second transferred from disk due to reactivationsduring the interval.



GBL_MEM_SWAPIN_BYTE_RATE_CUMThe average number of KBs per second transferred from disk due toreactivations over the cumulative collection time.


Process swapping has been replaced in HPUX 10.0 by a combination ofpaging and deactivation. Process deactivation occurs when the system isthrashing or when the amount of free memory falls below a critical level.The swapper then marks certain processes for deactivation and removes


242

them from the run queue. Pages within the associated memory regionsare reused or paged out by the memory management vhand process infavor of pages belonging to processes that are not deactivated. Unliketraditional process swapping, deactivated memory pages may or may notbe written out to the swap area, because a process could be reactivatedbefore the paging occurs.


GBL_MEM_SWAPIN_BYTE_RATE_HIGHThe highest number of KBs per second transferred from disk due toreactivations during any interval over the cumulative collection time.




GBL_MEM_SWAPIN_CUMThe number of reactivations over the cumulative collection time.


243









GBL_MEM_SWAPIN_RATEThe number of reactivations per second during the interval.



244








GBL_MEM_SWAPIN_RATE_CUMThe average number of reactivations per second over the cumulativecollection time.



To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swapareas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smaller


245

than HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process andreading in any pages that were actually paged out to the swap area whilethe process was deactivated.






GBL_MEM_SWAPIN_RATE_HIGHThe highest number of reactivations per second during any interval overthe cumulative collection time.






246





GBL_MEM_SWAPOUTThe number of deactivations during the interval.



This is the same as the “swap outs” value from the “vmstat -s” command.Remember that “vmstat -s” reports cumulative counts.

This metric can be compared to the “so” value from the “vmstat -S”command. The “so” value reports the number of processes swapped out(or deactivated) in HP-UX 10.0.

To determine the count (that is, GBL_MEM_SWAPOUT) for the currentinterval, subtract the previous value from the current value.

To determine the rate (that is, GBL_MEM_SWAPOUT_RATE) for thecurrent interval, subtract the previous value from the current value andthen divide by the length of the interval.



247

GBL_MEM_SWAPOUT_BYTEThe number of KBs (or MBs if specified) transferred out to disk due todeactivations during the interval.



GBL_MEM_SWAPOUT_BYTE_CUMThe number of KBs (or MBs if specified) transferred out to disk due todeactivations over the cumulative collection time.



To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swapareas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smallerthan HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now represent


248

bytes paged in as a result of reactivating a deactivated process andreading in any pages that were actually paged out to the swap area whilethe process was deactivated.

GBL_MEM_SWAPOUT_BYTE_RATEThe number of KBs per second transferred out to disk due to deactivationsduring the interval.



GBL_MEM_SWAPOUT_BYTE_RATE_CUMThe average number of KBs per second transferred out to disk due todeactivations over the cumulative collection time.



To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swap


249

areas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smallerthan HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process andreading in any pages that were actually paged out to the swap area whilethe process was deactivated.

GBL_MEM_SWAPOUT_BYTE_RATE_HIGHThe highest number of KBs per second transferred out to disk due todeactivations during any interval over the cumulative collection time.




GBL_MEM_SWAPOUT_CUMThe number of deactivations over the cumulative collection time.


Process swapping has been replaced in HPUX 10.0 by a combination ofpaging and deactivation. Process deactivation occurs when the system isthrashing or when the amount of free memory falls below a critical level.The swapper then marks certain processes for deactivation and removesthem from the run queue. Pages within the associated memory regionsare reused or paged out by the memory management vhand process in


250

favor of pages belonging to processes that are not deactivated. Unliketraditional process swapping, deactivated memory pages may or may notbe written out to the swap area, because a process could be reactivatedbefore the paging occurs.







GBL_MEM_SWAPOUT_RATEThe number of deactivations per second during the interval.


To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swapareas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smallerthan HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process and


251

reading in any pages that were actually paged out to the swap area whilethe process was deactivated.






GBL_MEM_SWAPOUT_RATE_CUMThe average number of deactivations per second over the cumulativecollection time.







252




GBL_MEM_SWAPOUT_RATE_HIGHThe highest number of deactivations per second during any interval overthe cumulative collection time.









253


GBL_MEM_SWAP_1_MIN_RATEThe number of deactivations/reactivations per minute during the interval.








GBL_MEM_SWAP_CUMThe total number of deactivations/reactivations over the cumulativecollection time.



254









GBL_MEM_SWAP_RATEThe number of deactivations/reactivations per second during the interval.



255








GBL_MEM_SWAP_RATE_CUMThe average number of deactivations/reactivations per second over thecumulative collection time.



To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swap


256

areas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smallerthan HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process andreading in any pages that were actually paged out to the swap area whilethe process was deactivated.






GBL_MEM_SWAP_RATE_HIGHThe highest number of deactivations/reactivations per second during anyinterval over the cumulative collection time.



To summarize, a process swap-out in HPUX 10.0 is a processdeactivation. A swap-in is a reactivation of a deactivated process. Swapmetrics that report swap-out bytes now represent bytes paged out to swapareas from deactivated regions. Because these pages are pushed outover time based on memory demands, these counts are much smallerthan HP-UX 9.x counts where the entire process was written to the swaparea when it was swapped-out. Likewise, swap-in bytes now representbytes paged in as a result of reactivating a deactivated process and


257

reading in any pages that were actually paged out to the swap area whilethe process was deactivated.






GBL_MEM_SYSThe amount of physical memory (in KBs unless otherwise specified) usedby the system (kernel) during the interval. System memory does notinclude the buffer cache.

On HP-UX 10.20 and 11.0, this metric does not include some kinds ofdynamically allocated kernel memory. This has always been reported inthe GBL_MEM_USER* metrics.

On HP-UX 11i and beyond, this metric includes some kinds of dynamicallyallocated kernel memory.

GBL_MEM_SYS_AND_CACHE_UTILThe percentage of physical memory used by the system (kernel) and thebuffer cache at the end of the interval.



GBL_MEM_SYS_UTILThe percentage of physical memory used by the system during theinterval.

System memory does not include the buffer cache.


258



GBL_MEM_USERThe amount of physical memory (in KBs unless otherwise specified)allocated to user code and data at the end of the interval. User memoryregions include code, heap, stack, and other data areas including sharedmemory. This does not include memory for buffer cache.

On HP-UX 10.20 and 11.0, this metric includes some kinds of dynamicallyallocated kernel memory.

On HP-UX 11i and beyond, this metric does not include some kinds ofdynamically allocated kernel memory. This is now reported in theGBL_MEM_SYS* metrics.

Large fluctuations in this metric can be caused by programs which allocatelarge amounts of memory and then either release the memory orterminate. A slow continual increase in this metric may indicate a programwith a memory leak.

GBL_MEM_USER_UTILThe percent of physical memory allocated to user code and data at theend of the interval. This metric shows the percent of memory owned byuser memory regions such as user code, heap, stack and other data areasincluding shared memory. This does not include memory for buffer cache.

On HP-UX 10.20 and 11.0, this metric includes some kinds of dynamicallyallocated kernel memory.

On HP-UX 11i and beyond, this metric does not include some kinds ofdynamically allocated kernel memory. This is now reported in theGBL_MEM_SYS* metrics.

Large fluctuations in this metric can be caused by programs which allocatelarge amounts of memory and then either release the memory orterminate. A slow continual increase in this metric may indicate a programwith a memory leak.

GBL_MEM_UTILThe percentage of physical memory in use during the interval. Thisincludes system memory (occupied by the kernel), buffer cache and usermemory.

This calculation is done using the byte values for physical memory andused memory, and is therefore more accurate than comparing the reportedkilobyte values for physical memory and used memory.


259

GBL_MEM_UTIL_CUMThe average percentage of physical memory in use over the cumulativecollection time. This includes system memory (occupied by the kernel),buffer cache and user memory.


GBL_MEM_UTIL_HIGHThe highest percentage of physical memory in use in any interval over thecumulative collection time.


GBL_MEM_VIRTThe total private virtual memory (in KBs unless otherwise specified) at theend of the interval. This is the sum of the virtual allocation of private dataand stack regions for all processes.

GBL_MEM_WAIT_PCTThe percentage of time processes or kernel threads were blocked on VM(waiting for virtual memory resources to become available) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on VM divided by the accumulated time that allprocesses or kernel threads were alive during the interval.




For example, the GBL_DISK_SUBSYSTEM_QUEUE values can be low,while the APP_DISK_SUBSYSTEM_WAIT_PCT values can be high. Inthis case, there are many processes on the system, but there are only avery small number of processes in the specific application that is being


260

examined and there is a high percentage of those few processes that areblocked on the disk I/O subsystem.

GBL_MEM_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on VM (waiting for virtual memory resources to becomeavailable) during the interval.

GBL_MSG_QUEUEThe average number of processes or kernel threads blocked on messages(waiting for their message queue calls to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on MESG (that is, messages) divided by the intervaltime.





GBL_MSG_WAIT_PCTThe percentage of time processes or kernel threads were blocked onmessages (waiting for their message queue calls to complete) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on MESG (that is, messages) divided by theaccumulated time that all processes or kernel threads were alive duringthe interval.




261



GBL_MSG_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on messages (waiting for their message queue calls tocomplete) during the interval.

GBL_NETWORK_SUBSYSTEM_QUEUEThe average number of processes or kernel threads blocked on thenetwork subsystem (waiting for their network activity to complete) duringthe interval. This is the sum of processes or kernel threads in the LAN,NFS, and RPC wait states. This does not include processes or kernelthreads blocked on SOCKT (that is, sockets) waits, as some processes orkernel threads sit idle in SOCKT waits for long periods.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (LAN + NFS + RPC) divided by the interval time.






262

GBL_NETWORK_SUBSYSTEM_WAIT_PCTThe percentage of time processes or kernel threads were blocked on thenetwork subsystem (waiting for their network activity to complete) duringthe interval. This is the sum of processes or kernel threads in the LAN,NFS, and RPC wait states. This does not include processes or kernelthreads blocked on SOCKT (that is, sockets) waits, as some processes orkernel threads sit idle in SOCKT waits for long periods.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on (LAN + NFS + RPC) divided by the accumulatedtime that all processes or kernel threads were alive during the interval.





GBL_NET_COLLISIONThe number of collisions that occurred on all network interfaces during theinterval. A rising rate of collisions versus outbound packets is anindication that the network is becoming increasingly congested. Thismetric does not include deferred packets.

For HP-UX 10.20 and earlier releases, this is the same as the sum of the“Coll” column from the “netstat -i” command for a network device. Seealso netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Single Collision Frames”, “Multiple Collision Frames”, “LateCollisions”, and “Excessive Collisions” values from the output of the“lanadmin” utility for the network interface. Remember that “lanadmin”reports cumulative counts. For this release and beyond, “netstat -i” showsnetwork activity on the logical level (IP) only.



263

GBL_NET_COLLISION_1_MIN_RATEThe number of collisions per minute on all network interfaces during theinterval.

Collisions occur on any busy network, but abnormal collision rates couldindicate a hardware or software problem.


GBL_NET_COLLISION_CUMThe number of collisions that occurred on all network interfaces over thecumulative collection time. A rising rate of collisions versus outboundpackets is an indication that the network is becoming increasinglycongested. This metric does not include deferred packets.


For HP-UX 10.20 and earlier releases, this is the same as the total sum ofthe “Coll” column from the “netstat -i” command for a network device. Seealso netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Single Collision Frames”, “Multiple Collision Frames”, “LateCollisions”, and “Excessive Collisions” values from the output of the“lanadmin” utility for the network interface. Remember that “lanadmin”reports cumulative counts. For this release and beyond, “netstat -i” showsnetwork activity on the logical level (IP) only.


GBL_NET_COLLISION_PCTThe percentage of collisions to total outbound packet attempts during theinterval. Outbound packet attempts include both successful packets andcollisions.

A rising rate of collisions versus outbound packets is an indication that thenetwork is becoming increasingly congested.

This metric does not currently include deferred packets.


GBL_NET_COLLISION_PCT_CUMThe percentage of collisions to total outbound packet attempts over thecumulative collection time. Outbound packet attempts include bothsuccessful packets and collisions.


264



This metric does not currently include deferred packets.


GBL_NET_COLLISION_RATEThe number of collisions per second on all network interfaces during theinterval.



GBL_NET_DEFERREDThe number of outbound deferred packets due to the network being in useduring the interval.


GBL_NET_DEFERRED_CUMThe number of outbound deferred packets due to the network being in useover the cumulative collection time.



GBL_NET_DEFERRED_PCTThe percentage of deferred packets to total outbound packet attemptsduring the interval. Outbound packet attempts include both packetssuccessfully transmitted and those that were deferred.



265

GBL_NET_DEFERRED_PCT_CUMThe percentage of deferred packets to total outbound packet attemptsover the cumulative collection time. Outbound packet attempts includeboth packets successfully transmitted and those that were deferred.



GBL_NET_DEFERRED_RATEThe number of deferred packets per second on all network interfacesduring the interval.


GBL_NET_DEFERRED_RATE_CUMThe number of deferred packets per second on all network interfaces overthe cumulative collection time.



GBL_NET_ERRORThe number of errors that occurred on all network interfaces during theinterval.

For HP-UX 10.20 and earlier releases, this is the same as the sum of“Ierrs” and “Oerrs” from the “netstat -i” command for a network device.See also netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Inbound Errors” and “Outbound Errors” values from the output of the“lanadmin” utility for the network interface. Remember that “lanadmin”reports cumulative counts. For this release and beyond, “netstat -i” showsnetwork activity on the logical level (IP) only.



266

GBL_NET_ERROR_1_MIN_RATEThe number of errors per minute on all network interfaces during theinterval. This rate should normally be zero or very small. A large errorrate can indicate a hardware or software problem.


GBL_NET_ERROR_CUMThe number of errors that occurred on all network interfaces over thecumulative collection time.


For HP-UX 10.20 and earlier releases, this is the same as the total sum of“Ierrs” and “Oerrs” from the “netstat -i” command for a network device.See also netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the total sumof the “Inbound Errors” and “Outbound Errors” values from the output ofthe “lanadmin” utility for the network interface. Remember that “lanadmin”reports cumulative counts. For this release and beyond, “netstat -i” showsnetwork activity on the logical level (IP) only.


GBL_NET_ERROR_RATEThe number of errors per second on all network interfaces during theinterval.


GBL_NET_IN_ERRORThe number of inbound errors that occurred on all network interfacesduring the interval.

A large number of errors may indicate a hardware problem on the network.

For HP-UX 10.20 and earlier releases, this is the same as the sum of“Ierrs” from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Inbound Errors” values from the output of the “lanadmin” utility for thenetwork interface. Remember that “lanadmin” reports cumulative counts.For this release and beyond, “netstat -i” shows network activity on thelogical level (IP) only.


267


GBL_NET_IN_ERROR_CUMThe number of inbound errors that occurred on all network interfaces overthe cumulative collection time.


A large number of errors may indicate a hardware problem on the network.

For HP-UX 10.20 and earlier releases, this is the same as the total sum of“Ierrs” from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the total sumof the “Inbound Errors” values from the output of the “lanadmin” utility forthe network interface. Remember that “lanadmin” reports cumulativecounts. For this release and beyond, “netstat -i” shows network activity onthe logical level (IP) only.


GBL_NET_IN_ERROR_PCTThe percentage of inbound network errors to total inbound packet attemptsduring the interval. Inbound packet attempts include both packetssuccessfully received and those that encountered errors.

A large number of errors may indicate a hardware problem on the network.The percentage of inbound errors to total packets attempted shouldremain low.


GBL_NET_IN_ERROR_PCT_CUMThe percentage of inbound network errors to total inbound packet attemptsover the cumulative collection time. Inbound packet attempts include bothpackets successfully received and those that encountered errors.




268


GBL_NET_IN_ERROR_RATEThe number of inbound errors per second on all network interfaces duringthe interval.



GBL_NET_IN_ERROR_RATE_CUMThe average number of inbound errors per second on all networkinterfaces over the cumulative collection time.



GBL_NET_IN_PACKETThe number of successful packets received through all network interfacesduring the interval. Successful packets are those that have beenprocessed without errors or collisions.

For HP-UX 10.20 and earlier releases, this is the same as the sum of the“Ipkts” column from the “netstat -i” command for a network device. Seealso netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Inbound Unicast Packets” and “Inbound Non-Unicast Packets” valuesfrom the output of the “lanadmin” utility for the network interface.Remember that “lanadmin” reports cumulative counts. For this releaseand beyond, “netstat -i” shows network activity on the logical level (IP)only.


GBL_NET_IN_PACKET_CUMThe number of successful packets received through all network interfacesover the cumulative collection time. Successful packets are those thathave been processed without errors or collisions.


269


For HP-UX 10.20 and earlier releases, this is the same as the total sum ofthe “Ipkts” column from the “netstat -i” command for a network device.See also netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the total sumof the “Inbound Unicast Packets” and “Inbound Non-Unicast Packets”values from the output of the “lanadmin” utility for the network interface.Remember that “lanadmin” reports cumulative counts. For this releaseand beyond, “netstat -i” shows network activity on the logical level (IP)only.


GBL_NET_IN_PACKET_RATEThe number of successful packets per second received through allnetwork interfaces during the interval. Successful packets are those thathave been processed without errors or collisions.


GBL_NET_IP_FRAGMENTS_RECEIVEDThe number of valid Ipv4 datagram fragments received by the host.

GBL_NET_IP_FWD_DATAGRAMSThe number of Ipv4 datagrams this host has forwarded. In other words,the number of Ipv4 datagrams for which this host has been used as arouter.

GBL_NET_IP_REASSEMBLY_REQUIREDThe number of Ipv4 datagram fragments sent to this host for local deliverywhich required reassembly before being given to the Upper LayerProtocol(s).

GBL_NET_OUTQUEUEThe sum of the outbound queue lengths for all network interfaces(BYNETIF_QUEUE). This metric is derived from the same source as theOutbound Queue Length shown in the lanadmin(1M) program.

For most interfaces, the outbound queue is usually zero. When the valueis non-zero over a period of time, the network may be experiencing a


270

bottleneck. Determine which network interface has a non-zero queue andcompare its traffic levels to normal. Also see if processes are blocking onnetwork wait states.



GBL_NET_OUT_ERRORThe number of outbound errors that occurred on all network interfacesduring the interval.

For HP-UX 10.20 and earlier releases, this is the same as the sum of“Oerrs” from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Outbound Errors” values from the output of the “lanadmin” utility forthe network interface. Remember that “lanadmin” reports cumulativecounts. For this release and beyond, “netstat -i” shows network activity onthe logical level (IP) only.


GBL_NET_OUT_ERROR_CUMThe number of outbound errors that occurred on all network interfacesover the cumulative collection time.


For HP-UX 10.20 and earlier releases, this is the same as the total sum of“Oerrs” from the “netstat -i” command for a network device. See alsonetstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the total sumof the “Outbound Errors” values from the output of the “lanadmin” utility forthe network interface. Remember that “lanadmin” reports cumulativecounts. For this release and beyond, “netstat -i” shows network activity onthe logical level (IP) only.


GBL_NET_OUT_ERROR_PCTThe percentage of outbound network errors to total outbound packetattempts during the interval. Outbound packet attempts include bothpackets successfully sent and those that encountered errors.


271

The percentage of outbound errors to total packets attempted to betransmitted should remain low.


GBL_NET_OUT_ERROR_PCT_CUMThe percentage of outbound network errors to total outbound packetattempts over the cumulative collection time. Outbound packet attemptsinclude both packets successfully sent and those that encountered errors.


The percentage of outbound errors to total packets attempted to betransmitted should remain low.


GBL_NET_OUT_ERROR_RATEThe number of outbound errors per second on all network interfacesduring the interval.


GBL_NET_OUT_ERROR_RATE_CUMThe number of outbound errors per second on all network interfaces overthe cumulative collection time.



GBL_NET_OUT_PACKETThe number of successful packets sent through all network interfacesduring the last interval. Successful packets are those that have beenprocessed without errors or collisions.

For HP-UX 10.20 and earlier releases, this is the same as the sum of the“Opkts” column from the “netstat -i” command for a network device. Seealso netstat(1).


272

For HP-UX 11.0 and beyond, this metric will be the same as the sum ofthe “Outbound Unicast Packets” and “Outbound Non-Unicast Packets”values from the output of the “lanadmin” utility for the network interface.Remember that “lanadmin” reports cumulative counts. For this releaseand beyond, “netstat -i” shows network activity on the logical level (IP)only.


GBL_NET_OUT_PACKET_CUMThe number of successful packets sent through all network interfaces overthe cumulative collection time. Successful packets are those that havebeen processed without errors or collisions.


For HP-UX 10.20 and earlier releases, this is the same as the total sum ofthe “Opkts” column from the “netstat -i” command for a network device.See also netstat(1).

For HP-UX 11.0 and beyond, this metric will be the same as the total sumof the “Outbound Unicast Packets” and “Outbound Non-Unicast Packets”values from the output of the “lanadmin” utility for the network interface.Remember that “lanadmin” reports cumulative counts. For this releaseand beyond, “netstat -i” shows network activity on the logical level (IP)only.


GBL_NET_OUT_PACKET_RATEThe number of successful packets per second sent through the networkinterfaces during the interval. Successful packets are those that havebeen processed without errors or collisions.


GBL_NET_PACKETThe total number of successful inbound and outbound packets for allnetwork interfaces during the interval. These are the packets that havebeen processed without errors or collisions.



273

GBL_NET_PACKET_RATEThe number of successful packets per second (both inbound andoutbound) for all network interfaces during the interval. Successfulpackets are those that have been processed without errors or collisions.


GBL_NFS_CALLThe number of NFS calls the local system has made as either a NFS clientor server during the interval.

This includes both successful and unsuccessful calls. Unsuccessful callsare those that cannot be completed due to resource limitations or LANpacket errors.

NFS calls include create, remove, rename, link, symlink, mkdir, rmdir,statfs, getattr, setattr, lookup, read, readdir, readlink, write, writecache, nulland root operations.

GBL_NFS_CALL_RATEThe number of NFS calls per second the system made as either a NFSclient or NFS server during the interval.

Each computer can operate as both a NFS server, and as an NFS client.

This metric includes both successful and unsuccessful calls. Unsuccessfulcalls are those that cannot be completed due to resource limitations orLAN packet errors.


GBL_NFS_CLIENT_BAD_CALLThe number of failed NFS client calls during the interval. Calls fail due tolack of system resources (lack of virtual memory) as well as networkerrors.

GBL_NFS_CLIENT_BAD_CALL_CUMThe number of failed NFS client calls over the cumulative collection time.Calls fail due to lack of system resources (lack of virtual memory) as wellas network errors.



274

GBL_NFS_CLIENT_BIODThe current number of biods running (both idle and active) at the end ofthe interval.

GBL_NFS_CLIENT_BYTEThe total number of KBs the local machine has sent or received as anNFS client during the interval. Each computer can operate as both anNFS server, and as a NFS client.

GBL_NFS_CLIENT_BYTE_CUMThe total number of KBs the local machine has sent or received as anNFS client over the cumulative collection time. Each computer canoperate as both a NFS server, and as a NFS client.


GBL_NFS_CLIENT_CALLThe number of NFS calls the local machine has processed as a NFS clientduring the interval. Calls are the system calls used to initiate physical NFSoperations. These calls are not always successful due to resourceconstraints or LAN errors, which means that the call rate should exceedthe IO rate. This metric includes both successful and unsuccessful calls.


GBL_NFS_CLIENT_CALL_CUMThe number of NFS calls the local machine has processed as a NFS clientover the cumulative collection time. Calls are the system calls used toinitiate physical NFS operations. These calls are not always successfuldue to resource constraints or LAN errors, which means that the call rateshould exceed the IO rate. This metric includes both successful andunsuccessful calls.




275

GBL_NFS_CLIENT_CALL_RATEThe number of NFS calls the local machine has processed as a NFS clientper second during the interval. Calls are the system call used to initiatephysical NFS operations. These calls are not always successful due toresource constraints or LAN errors, which means that the call rate shouldexceed the IO rate. This metric includes both successful and unsuccessfulcalls.


GBL_NFS_CLIENT_IDLE_BIODThe current number of biods inactive at the end of the interval. A value ofzero indicates a potential bottleneck for the NFS client.

GBL_NFS_CLIENT_IOThe number of NFS IOs the local machine has completed as an NFS clientduring the interval. This number represents physical IOs sent by the clientin contrast to a call which is an attempt to initiate these operations.


NFS IOs include reads and writes from successful calls to getattr, setattr,lookup, read, readdir, readlink, write, and writecache.

GBL_NFS_CLIENT_IO_CUMThe number of NFS IOs the local machine has completed as an NFS clientover the cumulative collection time. This number represents physical IOssent by the client in contrast to a call which is an attempt to initiate theseoperations. Each computer can operate as both a NFS server, and as anNFS client.



GBL_NFS_CLIENT_IO_PCTThe percentage of NFs IOs the local machine has completed as an NFSclient versus total NFS IOs completed during the interval. This numberrepresents physical IOs sent by the client in contrast to a call which is anattempt to initiate these operations.



276

A percentage greater than 50 indicates that this machine is acting more asa client. A percentage less than 50 indicates this machine is acting moreas a server for others.


GBL_NFS_CLIENT_IO_PCT_CUMThe percentage of NFS IOs the local machine has completed as an NFSclient versus total NFS IOs completed over the cumulative collection time.This number represents physical IOs sent by the client in contrast to a callwhich is an attempt to initiate these operations. Each computer canoperate as both a NFS server, and as an NFS client.


A percentage greater than 50 indicates that this machine is acting more asa client. A percentage less than 50 indicates this machine is acting moreas a server for others.


GBL_NFS_CLIENT_IO_RATEThe number of NFS IOs per second the local machine has completed asan NFS client during the interval. This number represents physical IOssent by the client in contrast to a call which is an attempt to initiate theseoperations.



GBL_NFS_CLIENT_IO_RATE_CUMThe number of NFS IOs per second the local machine has completed asan NFS client over the cumulative collection time. This number representsphysical IOs sent by the client in contrast to a call which is an attempt toinitiate these operations. Each computer can operate as both a NFSserver, and as an NFS client.




277

GBL_NFS_CLIENT_PHYS_TIMEThe time, in seconds, spent to service all NFS operations as a NFS clientduring the last interval. This is measured from the time the operation getsonto the physical network until the time a reply is received from thenetwork. In other words, this is the “service time” less the local machine'ssoftware overhead.

GBL_NFS_CLIENT_PHYS_TIME_CUMThe time, in seconds, spent to service all NFS operations as a NFS clientover the cumulative collection time. This is measured from the time theoperation gets onto the physical network until the time a reply is receivedfrom the network. In other words, this is the “service time” less the localmachine's software overhead.


GBL_NFS_CLIENT_READ_BYTE_RATEThe number of KBs per second the system received as an NFS clientdoing read operations during the interval.



GBL_NFS_CLIENT_READ_BYTE_RATE_CUMThe average number of KBs per second the system received as an NFSclient doing read operations over the cumulative collection time.




GBL_NFS_CLIENT_READ_RATEThe number of NFS “read” operations per second the system generatedas an NFS client during the interval.


278



GBL_NFS_CLIENT_READ_RATE_CUMThe average number of NFS “read” operations per second the systemgenerated as an NFS client over the cumulative collection time.




GBL_NFS_CLIENT_SERVICE_QUEUEThe number of pending NFS client operations during the interval. Thisvalue increases as the service time increases and/or as the rate of clientrequests increases.

GBL_NFS_CLIENT_SERVICE_QUEUE_CUMThe average number of pending NFS client operations per interval overthe cumulative collection time. Queue length increases as the service timeincreases and/or as the rate of client requests increases.


GBL_NFS_CLIENT_SERVICE_TIMEThe time, in seconds, spent to service all NFS operations as a NFS clientduring the last interval. This is the time from the point that the clientoriginates the requests to the point replies are received including IObuffering, NFS and network software layer delays, physical networklatency, and NFS server service time. It is not a measure of the averageresponse time per NFS request. This can be thought of as the round-triptime for all NFS requests made during the interval.


279

GBL_NFS_CLIENT_SERVICE_TIME_CUMThe time, in seconds, spent to service all NFS operations as a NFS clientover the cumulative collection time. This is the time from the point that theclient originates the request to the point a reply is received including IObuffering, NFS and network software layer delays, physical networklatency, and NFS server service time. It is not a measure of the averageresponse time per nfs request. This can be thought of as the round-triptime for all nfs requests.


GBL_NFS_CLIENT_WRITE_BYTE_RATEThe number of KBs per second the system sent over the network as anNFS client doing write operations during the interval.



GBL_NFS_CLIENT_WRITE_BYTE_RATE_CUMThe average number of KBs per second the system sent over the networkas an NFS client doing write operations over the cumulative collectiontime.




GBL_NFS_CLIENT_WRITE_RATEThe number of NFS “write” operations per second the system generatedas an NFS client during the interval.




280

GBL_NFS_CLIENT_WRITE_RATE_CUMThe average number of NFS “write” operations per second the systemgenerated as an NFS client over the cumulative collection time.




GBL_NFS_LOGL_READThe number of logical reads made to NFS disks by the local machine as aNFS client during the interval.

Each computer can operate as both a NFS server, and as an NFS client.For this metric the local machine is acting as a NFS client (that is, thedisks are remote) since if it were acting as a server the logical diskrequests would be going to local disks. These logical requests do notnecessarily result in a physical IO request across the NFS link.

GBL_NFS_LOGL_READ_BYTEThe number of KBs transferred through logical reads to NFS disks by thelocal machine during the interval. Note that these are transfers by readcalls, not physical IO.

GBL_NFS_LOGL_READ_BYTE_CUMThe number of KBs transferred through logical reads to NFS disks by thelocal machine over the cumulative collection time. Note that these aretransfers by read calls, not physical IO.


GBL_NFS_LOGL_READ_CUMThe total number of logical reads made to NFS disks by the local machineas a NFS client over the cumulative collection time.



281

Each computer can operate as both a NFS server, and as an NFS client.For this metric the local machine is acting as an NFS client (the disks areremote) since if it were acting as a server the logical disk requests wouldbe going to local disks. These logical requests do not necessarily result ina physical IO request across the NFS link.

GBL_NFS_LOGL_READ_PCTThe percentage of logical reads to total logical reads and writes to NFSdisks by the local machine during the interval.

GBL_NFS_LOGL_READ_PCT_CUMThe average percentage of logical reads to total logical reads and writes toNFS disks by the local machine over the cumulative collection time.


GBL_NFS_LOGL_READ_RATEThe number of logical reads per second made to NFS disks by the localmachine during the interval.

GBL_NFS_LOGL_READ_RATE_CUMThe average number of logical reads per second made to NFS disks bythe local machine over the cumulative collection time.


GBL_NFS_LOGL_WRITEThe number of logical writes made to NFS disks by the local machineduring the interval.

Each computer can operate as both a NFS server, and as a NFS client.For this metric the local machine is acting as an NFS client (the disks areremote) since if it were acting as a server the logical disk requests wouldbe going to local disks. These logical requests do not necessarily result ina physical IO request across the NFS link.


282

GBL_NFS_LOGL_WRITE_BYTEThe number of KBs transferred through logical writes to NFS disks by thelocal machine during the interval. Note that these are transfers by writecalls, not physical IO.

GBL_NFS_LOGL_WRITE_BYTE_CUMThe number of KBs transferred through logical writes to NFS disks by thelocal machine over the cumulative collection time. Note that these aretransfers by write calls, not physical IO.


GBL_NFS_LOGL_WRITE_CUMThe total number of logical writes made to NFS disks by the local machineover the cumulative collection time.


Each computer can operate as both a NFS server, and as a NFS client.For this metric the local machine is acting as an NFS client (the disks areremote) since if it were acting as a server the logical disk requests wouldbe going to local disks. These logical requests do not necessarily result ina physical IO request across the NFS link.

GBL_NFS_LOGL_WRITE_PCTThe percentage of logical writes to total logical reads and writes to NFSdisks by the local machine during the interval.

GBL_NFS_LOGL_WRITE_PCT_CUMThe average percentage of logical writes to total logical IO to NFS disks bythe local machine over the cumulative collection time.


GBL_NFS_LOGL_WRITE_RATEThe number of logical writes per second made to NFS disks by the localmachine during the interval.


283

GBL_NFS_LOGL_WRITE_RATE_CUMThe average number of logical writes per second made to NFS disks bythe local machine over the cumulative collection time.


GBL_NFS_QUEUEThe average number of processes or kernel threads blocked on NFS(waiting for their network file system IO to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on NFS divided by the interval time.





GBL_NFS_SERVER_BAD_CALLThe number of failed NFS server calls during the interval. Calls fail due tolack of system resources (lack of virtual memory) as well as networkerrors.

GBL_NFS_SERVER_BAD_CALL_CUMThe number of failed NFS server calls over the cumulative collection time.Calls fail due to lack of system resources (lack of virtual memory) as wellas network errors.



284


GBL_NFS_SERVER_BYTEThe number of KBs the local machine has processed as a NFS serverduring the interval. Each computer can operate as both an NFS server,and as a NFS client.

GBL_NFS_SERVER_BYTE_CUMThe number of KBs the local machine has processed as a NFS serverover the cumulative collection time. Each computer can operate as bothan NFS server, and as a NFS client.


GBL_NFS_SERVER_CALLThe number of NFS calls the local machine has processed as a NFSserver during the interval.

Calls are the system calls used to initiate physical NFS operations. Thesecalls are not always successful due to resource constraints or LAN errors,which means that the call rate could exceed the IO rate. This metricincludes both successful and unsuccessful calls.


GBL_NFS_SERVER_CALL_CUMThe number of NFS calls the local machine has processed as a NFSserver over the cumulative collection time.





285

GBL_NFS_SERVER_CALL_RATEThe number of NFS calls the local machine has processed per second asa NFS server during the interval.



GBL_NFS_SERVER_IOThe number of NFS IOs the local machine has completed as an NFSserver during the interval. This number represents physical IOs receivedby the server in contrast to a call which is an attempt to initiate theseoperations.



GBL_NFS_SERVER_IO_CUMThe number of NFS IOs the local machine has completed as an NFSserver over the cumulative collection time. This number representsphysical IOs received by the server n contrast to a call which is an attemptto initiate these operations. Each computer can operate as both a NFSserver, and as an NFS client.



GBL_NFS_SERVER_IO_PCTThe percentage of NFS IOs the local machine has completed as an NFSserver versus total NFS IOs completed during the interval. This numberrepresents physical IOs received by the server in contrast to a call which isan attempt to initiate these operations.


A percentage greater than 50 indicates that this machine is acting more asa server for others. A percentage less than 50 indicates this machine isacting more as a client.


286


GBL_NFS_SERVER_IO_PCT_CUMThe percentage of NFs IOs the local machine has completed as an NFSserver versus total NFS IOs completed over the cumulative collection time.This number represents physical IOs received by the server in contrast toa call which is an attempt to initiate these operations. Each computer canoperate as both a NFS server, and as an NFS client.


A percentage greater than 50 indicates that this machine is acting more asa server for others. A percentage less than 50 indicates this machine isacting more as a client.


GBL_NFS_SERVER_IO_RATEThe number of NFS IOs per second the local machine has completed asan NFS server during the interval. This number represents physical IOsreceived by the server in contrast to a call which is an attempt to initiatethese operations.



GBL_NFS_SERVER_IO_RATE_CUMThe number of NFS IOs per second the local machine has completed asan NFS server over the cumulative collection time. This numberrepresents physical IOs received by the server in contrast to a call which isan attempt to initiate these operations. Each computer can operate asboth a NFS server, and as an NFS client.




287

GBL_NFS_SERVER_READ_BYTE_RATEThe number of KBs per second the system sent as a NFS serverresponding to NFS read operations from client nodes during the interval.



GBL_NFS_SERVER_READ_BYTE_RATE_CUMThe average number of KBs per second the system sent as an NFS serverresponding to NFS read operations from client nodes over the cumulativecollection time.




GBL_NFS_SERVER_READ_RATEThe number of NFS “read” operations per second the system processedas an NFS server during the interval.



GBL_NFS_SERVER_READ_RATE_CUMThe average number of NFS “read” operations per second the systemprocessed as an NFS server over the cumulative collection time.





288

GBL_NFS_SERVER_SERVICE_TIMEThe time, in seconds, spent for the NFS server to process the client'soperations during the interval. This includes all of the time from the pointthat the operations are received to the point where a reply is sent back tothe client, which includes software overhead and any local disk IOs. Thisis not an average service time per operation; it is the total service time forall operations processed during the interval.

GBL_NFS_SERVER_SERVICE_TIME_CUMThe time, in seconds, spent for the NFS server to process the client'soperations over the cumulative collection time. This includes all of thetime from the point that the operations are received to the point where areply is sent back to the client, which includes software overhead and anylocal disk IOs. This is not an average service time per operation; it is thetotal service time for all operations processed.


GBL_NFS_SERVER_WRITE_BYTE_RATEThe number of KBs per second the system received over the network asan NFS server performing write operations for client nodes during theinterval.



GBL_NFS_SERVER_WRITE_BYTE_RATE_CUMThe average number of KBs per second the system received over thenetwork as an NFS server performing write operations for client nodesover the cumulative collection time.





289

GBL_NFS_SERVER_WRITE_RATEThe number of NFS “write” operations per second the system processedas an NFS server during the interval.



GBL_NFS_SERVER_WRITE_RATE_CUMThe average number of NFS “write” operations per second the systemprocessed as an NFS server over the cumulative collection time.




GBL_NFS_WAIT_PCTThe percentage of time processes or kernel threads were blocked on NFS(waiting for their network file system IO to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on NFS divided by the accumulated time that allprocesses or kernel threads were alive during the interval.






290

GBL_NFS_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on NFS (waiting for their network file system IO to complete)during the interval.

GBL_NODENAMEThe name of the computer as returned by the command “uname -n” (thatis, the string returned from the “hostname” program).

GBL_NUM_APPThe number of applications defined in the parm file plus one (for “other”).

The application called “other” captures all other processes not defined inthe parm file.

You can define up to 128 applications.

GBL_NUM_APP_PRMThe number of PRM groups configured - 1 per PRM Group ID. HP-UX10.0 supports up to 16 unique PRM Groups. For HP-UX 10.20 andbeyond, this value has been increased to 64.

GBL_NUM_CPUThe number of CPUs physically on the system.

The sar(1M) command allows you to check the status of the system CPUs.

GBL_NUM_DISKThe number of disks on the system.

This is a count of the number of disks on the system that have ever hadactivity over the cumulative collection time.


“Disk” refers to a physical drive (that is, “spindle”), not a partition on adrive (unless the partition occupies the entire physical disk).


291

GBL_NUM_NETWORKThe number of Local Area Network (LAN) interfaces on the system. Thisincludes the loopback interface. The “netstat -i” command also displaysthe list of network interfaces on the system.

GBL_NUM_SWAPThe number of configured swap areas.

GBL_NUM_TTThe number of unique Transaction Tracker (TT) transactions that havebeen registered on this system.

GBL_NUM_USERThe information for this metric comes from the utmp file which is updatedby the login command. For more information, read the man page for utmp.Some applications may create users on the system without using login andupdating the utmp file. These users are not reflected in this count.

This metric can be a general indicator of system usage. In a networkedenvironment, however, users may maintain inactive logins on severalsystems.

The number of users logged in at the time of the interval sample. This isthe same as the command “who|wc -l”, which may exceed the value in thekernel parameter “maxusers”.

Levels of remote users that are close to the configured maximum value(npty) may cause problems because login attempts can fail when that limitis reached.

GBL_NUM_VGThe number of available volume groups.

GBL_OSKERNELTYPEThis indicates the word size of the current kernel on the system. Somehardware can load the 64-bit kernel or the 32-bit kernel.


GBL_OSKERNELTYPE_INTThis indicates the word size of the current kernel on the system. Somehardware can load the 64-bit kernel or the 32-bit kernel.



292

GBL_OSNAMEA string representing the name of the operating system. This is the sameas the output from the “uname -s” command. For example, “HP-UX”.

GBL_OSRELEASEThe current release of the operating system. This is the same as theoutput from the “uname -r” command.

GBL_OSVERSIONA string representing the version of the operating system. This is thesame as the output from “uname -v”.

This is an alphabetic code indicating the number of users specified by theoperating system license. The mapping is documented under the“uname(2)” command.

For the Series 700/800:

A => two-user system

B => 16-user system

C => 32-user system

D => 64-user system

E => 8-user system

U => unlimited-users system

GBL_OTHER_IO_QUEUEThe average number of processes or kernel threads blocked on “other IO”during the interval. “Other IO” includes all IO directed at a device(connected to the local computer) which is not a terminal or LAN.Examples of “other IO” devices are local printers, tapes, instruments, anddisks. Time waiting for character (raw) IO to disks is included in thismeasurement. Time waiting for file system buffered IO to disks willtypically been seen as IO or CACHE wait. Time waiting for IO to NFSdisks is reported as NFS wait.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on other IO divided by the interval time.





293



GBL_OTHER_IO_WAIT_PCTThe percentage of time processes or kernel threads were blocked on“other IO” during the interval. “Other IO” includes all IO directed at adevice (connected to the local computer) which is not a terminal or LAN.Examples of “other IO” devices are local printers, tapes, instruments, anddisks. Time waiting for character (raw) IO to disks is included in thismeasurement. Time waiting for file system buffered IO to disks willtypically been seen as IO or CACHE wait. Time waiting for IO to NFSdisks is reported as NFS wait.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on other IO divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





GBL_OTHER_IO_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on “other IO” during the interval. “Other IO” includes all IOdirected at a device (connected to the local computer) which is not aterminal or LAN. Examples of “other IO” devices are local printers, tapes,instruments, and disks. Time waiting for character (raw) IO to disks isincluded in this measurement. Time waiting for file system buffered IO todisks will typically been seen as IO or CACHE wait. Time waiting for IO toNFS disks is reported as NFS wait.


294

GBL_OTHER_QUEUEThe average number of processes or kernel threads blocked on other(unknown) activities during the interval. This includes processes or kernelthreads that were started and subsequently suspended before themidaemon was started and have not been resumed, or the block state isunknown.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on OTHER divided by the interval time.





GBL_OTHER_WAIT_PCTThe percentage of time processes or kernel threads were blocked on other(unknown) activities during the interval. This includes processes or kernelthreads that were started and subsequently suspended before themidaemon was started and have not been resumed, or the block state isunknown.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on OTHER divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





295


GBL_OTHER_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on other (unknown) activities during the interval. Thisincludes processes or kernel threads that were started and subsequentlysuspended before the midaemon was started and have not been resumed,or the block state is unknown.

GBL_PIPE_QUEUEThe average number of processes or kernel threads blocked on PIPE(waiting for pipe communication to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on PIPE divided by the interval time.





GBL_PIPE_WAIT_PCTThe percentage of time processes or kernel threads were blocked on PIPE(waiting for pipe communication to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on PIPE divided by the accumulated time that allprocesses or kernel threads were alive during the interval.



296




GBL_PIPE_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on PIPE (waiting for pipe communication to complete) duringthe interval.

GBL_PRI_QUEUEThe average number of processes or kernel threads blocked on PRI(waiting for their priority to become high enough to get the CPU) during theinterval.

To determine if the CPU is a bottleneck, compare this metric withGBL_CPU_TOTAL_UTIL. If GBL_CPU_TOTAL_UTIL is near 100 percentand GBL_PRI_QUEUE is greater than three, there is a high probability ofa CPU bottleneck.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on PRI divided by the interval time.


GBL_RUN_QUEUE = 1.0

GBL_PRI_QUEUE = 0.1

GBL_CPU_QUEUE = 1.0


GBL_RUN_QUEUE = 2.0

GBL_PRI_QUEUE = 8.0

GBL_CPU_QUEUE = 9.0


297



Note that if the value for GBL_PRI_QUEUE greatly exceeds the value forGBL_RUN_QUEUE, this may be a side-effect of the measurementinterface having lost trace data. In this case, check the value of theGBL_LOST_MI_TRACE_BUFFERS metric. If there has been buffer loss,you can correct the value of GBL_PRI_QUEUE by restarting themidaemon and the performance tools. You can use the/opt/perf/bin/midaemon -T command to force immediate shutdown of themeasurement interface.





GBL_PRI_WAIT_PCTThe percentage of time processes or kernel threads were blocked on PRI(waiting for their priority to become high enough to get the CPU) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on PRI divided by the accumulated time that allprocesses or kernel threads were alive during the interval.



298




GBL_PRI_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on PRI (waiting for their priority to become high enough toget the CPU) during the interval.

GBL_PRM_MEM_UTILThe total percent of memory used by processes within the PRM groupsduring the interval. This does not include system processes (processesattached to PRM group 0).

GBL_PROC_RUN_TIMEThe average run time, in seconds, for processes that terminated during theinterval.

GBL_PROC_SAMPLEThe number of process data samples that have been averaged into globalmetrics that are based on process samples, such asGBL_ACTIVE_PROC.

GBL_RPC_QUEUEThe average number of processes or kernel threads blocked on RPC(waiting for their remote procedure calls to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on RPC divided by the interval time.



299




GBL_RPC_WAIT_PCTThe percentage of time processes or kernel threads were blocked on RPC(waiting for their remote procedure calls to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on RPC divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





GBL_RPC_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on RPC (waiting for their remote procedure calls tocomplete) during the interval.


300

GBL_RUN_QUEUEThe average number of “runnable” processes or kernel threads over allprocessors during the interval. The value shown represents the averageof the 1-minute load averages for all processors.

This metric is derived from a kernel variable (avenrun) which is calculatedby summing the number of runnable processes or kernel threads for eachprocessor and averaging the samples over the last minute. Processes orkernel threads marked “runnable” include:





sample




11.0)





GBL_RUN_QUEUE = 1.0

GBL_PRI_QUEUE = 0.1

GBL_CPU_QUEUE = 1.0


GBL_RUN_QUEUE = 2.0

GBL_PRI_QUEUE = 8.0

GBL_CPU_QUEUE = 9.0


This illustrates that the run queue is the average of the 1-minute loadaverages for all processors; the pri queue is the number of processes or


301

kernel threads that are blocked on “PRI” (priority); and the cpu queue isthe number of processes or kernel threads in the cpu queue that are readyto run, including the processes or kernel threads using the CPU.

To determine if the CPU is a bottleneck, examine this metric along withGBL_CPU_TOTAL_UTIL and GBL_PRI_QUEUE. IfGBL_CPU_TOTAL_UTIL is near 100 percent and GBL_PRI_QUEUE isgreater than three, there is a high probability of a CPU bottleneck.

GBL_RUN_QUEUE_CUMThe average number of “runnable” processes or kernel threads over allprocessors over the cumulative collection time. The value shownrepresents the average of the 1-minute load averages for all processors.


In this case, this metric is a cumulative average of data that was collectedas an average. This metric is derived from GBL_RUN_QUEUE.

This metric is derived from a kernel variable (avenrun) which is calculatedby summing the number of runnable processes or kernel threads for eachprocessor and averaging the samples over the last minute. Processes orkernel threads marked “runnable” include:





sample




11.0)





GBL_RUN_QUEUE = 1.0

GBL_PRI_QUEUE = 0.1


302

GBL_CPU_QUEUE = 1.0


GBL_RUN_QUEUE = 2.0

GBL_PRI_QUEUE = 8.0

GBL_CPU_QUEUE = 9.0



GBL_RUN_QUEUE_HIGHThe highest average number of “runnable” processes or kernel threadsover all processors over the cumulative collection time. The value shownrepresents the average of the 1-minute load averages for all processors.


GBL_SAMPLEThe number of data samples (intervals) that have occurred over thecumulative collection time.


GBL_SEM_QUEUEThe average number of processes or kernel threads blocked onsemaphores (waiting for their semaphore operations to complete) duringthe interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on PRI (that is, priority) divided by the interval time.



303




GBL_SEM_WAIT_PCTThe percentage of time processes or kernel threads were blocked onsemaphores (waiting on a semaphore operation) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SEM (that is, semaphores) divided by theaccumulated time that all processes or kernel threads were alive duringthe interval.





GBL_SEM_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on semaphores (waiting for their semaphore operations tocomplete) during the interval.


304

GBL_SERIALNOThe ID number of the computer as returned by the command “uname -i”.If this value is not available, an empty string is returned.

GBL_SLEEP_QUEUEThe average number of processes or kernel threads blocked on SLEEP(waiting to awaken from sleep system calls) during the interval. A processor kernel thread enters the SLEEP state by putting itself to sleep usingsystem calls such as sleep, wait, pause, sigpause, sigsuspend, poll andselect.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SLEEP divided by the interval time.





GBL_SLEEP_WAIT_PCTThe percentage of time processes or kernel threads were blocked onSLEEP (waiting to awaken from sleep system calls) during the interval. Aprocess or kernel thread enters the SLEEP state by putting itself to sleepusing system calls such as sleep, wait, pause, sigpause, sigsuspend, polland select.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SLEEP divided by the accumulated time that allprocesses or kernel threads were alive during the interval.



No direct comparison is reasonable with the Application WAIT PCTmetrics since they represent percentages within the context of a specific


305

application and cannot be summed or compared with global values easily.In addition, the sum of each Application WAIT PCT for all applications willnot equal 100% since these values will vary greatly depending on thenumber of processes or kernel threads in each application.


GBL_SLEEP_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on SLEEP (waiting to awaken from sleep system calls)during the interval. A process or kernel thread enters the SLEEP state byputting itself to sleep using system calls such as sleep, wait, pause,sigpause, sigsuspend, poll and select.

GBL_SOCKET_QUEUEThe average number of processes or kernel threads blocked on sockets(waiting for their IO to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SOCKT (that is, sockets) divided by the intervaltime.





GBL_SOCKET_WAIT_PCTThe percentage of time processes or kernel threads were blocked onsockets (waiting for their IO to complete) during the interval.


306

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SOCKT (that is, sockets) divided by theaccumulated time that all processes or threads were alive during theinterval.





GBL_SOCKET_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on sockets (waiting for their IO to complete) during theinterval.

GBL_STARTDATEThe date that the collector started.

GBL_STARTED_PROCThe number of processes that started during the interval.

GBL_STARTTIMEThe time of day that the collector started.

GBL_STATDATEThe date at the end of the interval, based on local time.


307

GBL_STATTIMEAn ASCII string representing the time at the end of the interval, based onlocal time.

GBL_STREAM_QUEUEThe average number of processes or kernel threads blocked on streamsIO (waiting for a streams IO operation to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on STRMS (that is, streams IO) divided by theinterval time.






GBL_STREAM_WAIT_PCTThe percentage of time processes or kernel threads were blocked onstreams IO (waiting for a streams IO operation to complete) during theinterval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on STRMS (that is, streams IO) divided by theaccumulated time that all processes or kernel threads were alive duringthe interval.



No direct comparison is reasonable with the Application WAIT PCTmetrics since they represent percentages within the context of a specificapplication and cannot be summed or compared with global values easily.


308

In addition, the sum of each Application WAIT PCT for all applications willnot equal 100% since these values will vary greatly depending on thenumber of processes or kernel threads in each application.



GBL_STREAM_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on streams IO (waiting for a streams IO operation tocomplete) during the interval.


GBL_SWAP_RESERVED_ONLY_UTILThe percentage of available swap space reserved (for currently runningprograms), but not yet used.

Swap space must be reserved (but not allocated) before virtual memorycan be created. Swap space locations are actually assigned (used) whena page is actually written to disk.

When compared to the “swapinfo -mt” command results, this is calculatedas:

Util = ((USED: reserve)

/ (AVAIL: total)) * 100


GBL_SWAP_SPACE_AVAILThe total amount of potential swap space, in MB. This is the sum of thedevice swap areas enabled by the swapon command, the allocated size ofany file system swap areas, and the allocated size of pseudo swap inmemory if enabled.

Note that this is potential swap space. Since swap is allocated in fixed(SWCHUNK) sizes, not all of this space may actually be usable. Forexample, on a 61MB disk using 2 MB swap size allocations, 1 MB remainsunusable and is considered wasted space.

This is the same as (AVAIL: total) as reported by the “swapinfo -mt”command.


309


GBL_SWAP_SPACE_AVAIL_KBThe total amount of potential swap space, in KB.

This is the sum of the device swap areas enabled by the swaponcommand, the allocated size of any file system swap areas, and theallocated size of pseudo swap in memory if enabled.

Note that this is potential swap space. Since swap is allocated in fixed(SWCHUNK) sizes, not all of this space may actually be usable. Forexample, on a 61MB disk using 2 MB swap size allocations, 1 MB remainsunusable and is considered wasted space.

This is the same as (AVAIL: total) as reported by the “swapinfo -t”command.


GBL_SWAP_SPACE_DEVICE_UTILThe percentage of device swap space currently in use of the total swapspace available. This does not include file system or remote swap space.

Note that available swap is only potential swap space. Since swap isallocated in fixed (SWCHUNK) sizes, not all of this space may actually beusable. For example, on a 61 MB disk using 2 MB swap size allocations,1 MB remains unusable and is considered wasted space. Consequently,100 percent utilization on a single device is not always obtainable.

The wasted swap space, and the remainder of allocated SWCHUNKs thathave not been used is what is reported in the hold field of the/usr/sbin/swapinfo command.


Util = ((USED: dev) sum



GBL_SWAP_SPACE_FS_UTILThe percentage file system swap space currently in use of the total swapspace available. This includes both local and NFS file system swap.

Since file system swap is dynamic (it grows in SWCHUNK sizes asneeded and is not bounded as device swap is), this number fluctuates asmore swap is allocated.


310


Util = ((USED: fs) sum



GBL_SWAP_SPACE_RESERVEDThe amount of swap space (in MB) reserved for the swapping and pagingof programs currently executing. Process pages swapped include data(heap and stack pages), bss (data uninitialized at the beginning of processexecution), and the process user area (uarea). Shared memory regionsalso require the reservation of swap space.

Swap space is reserved (by decrementing a counter) when virtual memoryfor a program is created, but swap is only used when a page or swap todisk is actually done or the page is locked in memory if swapping tomemory is enabled. Virtual memory cannot be created if swap spacecannot be reserved.

This is the same as (USED: total) as reported by the “swapinfo -mt”command.


GBL_SWAP_SPACE_RESERVED_UTILThis is the percentage of available swap space currently reserved forrunning processes.

Reserved utilization = (amount of swap space reserved / amount of swapspace available) * 100

On HP-UX systems, swap space must be reserved (but not allocated)before virtual memory can be created. If all of available swap is reserved,then no new processes or virtual memory can be created. Swap spacelocations are actually assigned (used) when a page is actually written todisk.



Util = ((USED: total)



311


GBL_SWAP_SPACE_USEDThe amount of swap space (in MB) used. “Used” indicates written to disk(or locked in memory), rather than reserved. swap space is allocated inSWCHUNK size increments (defaulting to 2 MB) which means that not allswap space may be used. For example, on a 61 MB disk using 2 MBswap size allocations, 1 MB remains unusable and is considered wastedspace.

This is the same as (USED: total - reserve) as reported by the “swapinfo -mt” command.


GBL_SWAP_SPACE_USED_UTILThe percentage of swap space currently in use (has memory belonging toprocesses paged or swapped out onto it).

“Used” indicates written to disk or locked in memory, rather than reserved.Swap space is allocated in SWCHUNK size increments (defaulting to 2MB) which means that not all swap space may be used. For example, ona 61 MB disk using 2 MB swap size allocations, 1 MB remains unusableand is considered wasted space.


Util = ((USED: total - reserve)



GBL_SWAP_SPACE_UTILThe percent of swap space available that was reserved by runningprocesses in the interval.

On HP-UX systems, swap space must be reserved (but not allocated)before virtual memory can be created. If all of available swap is reserved,then no new processes or virtual memory can be created. Swap spacelocations are actually assigned (used) when a page is actually written todisk or locked in memory (pseudo swap in memory).

Note that available swap is only potential swap space. Since swap isallocated in fixed (SWCHUNK) sizes, not all of this space may actually beusable. For example, on 61 MB disk using 2 MB swap size allocations, 1MB remains unusable and is considered wasted space. Consequently,100 percent utilization on a single device is not always obtainable.


312

This is the same as (PCT USED: total) as reported by the “swapinfo -mt”command.

This metric is a measure of capacity rather than performance. As thismetric nears 100 percent, processes are not able to allocate any morememory and new processes may not be able to run. Very low swaputilization values may indicate that too much area has been allocated toswap, and better use of disk space could be made by reallocating someswap partitions to be user filesystems.


GBL_SWAP_SPACE_UTIL_CUMThe average percentage of available swap space currently in use (hasmemory belonging to processes paged or swapped out on it) over thecumulative collection time.




GBL_SWAP_SPACE_UTIL_HIGHThe highest average percentage of available swap space currently in use(has memory belonging to processes paged or swapped out on it) in anyinterval over the cumulative collection time.




GBL_SYSCALL_RATEThe average number of system calls per second during the interval.


313

High system call rates are normal on busy systems, especially with IOintensive applications. Abnormally high system call rates may indicateproblems such as a “hung” terminal that is stuck in a loop generating readsystem calls.

On HP-UX, system call rates affect the overhead of the midaemon.

Due to the system call instrumentation, the fork and vfork system calls aredouble counted. In the case of fork and vfork, one process starts thesystem call, but two processes exit.

Lightweight system calls, such as umask, do not show up in theGlancePlus System Calls display, but will get added to the global systemcall rates. If a process is being traced (debugged) using standarddebugging tools (such as adb or xdb), all system calls used by thatprocess will show up in the System Calls display while being traced.

Compare this metric to GBL_DISK_LOGL_IO_RATE to see if high systemcallrates correspond to high disk IO. GBL_CPU_SYSCALL_UTIL showsthe CPU utilization due to processing system calls.

GBL_SYSCALL_RATE_CUMThe average number of system calls per second over the cumulativecollection time.




GBL_SYSCALL_RATE_HIGHThe highest number of system calls per second during any interval overthe cumulative collection time.



Lightweight system calls, such as umask, do not show up in theGlancePlus System Calls display, but will get added to the global systemcall rates. If a process is being traced (debugged) using standard


314

debugging tools (such as adb or xdb), all system calls used by thatprocess will show up in the System Calls display while being traced.

GBL_SYSTEM_IDThe system ID as defined in the parm file which specifies the name of thesystem that the data was collected on. If the ID is not specified in theparm file, the system ID defaults to the nodename displayed when youenter the command “uname -n”.

GBL_SYSTEM_TYPEThe model of system. For example, “HP9000”.

GBL_SYSTEM_UPTIME_HOURSThe time, in hours, since the last system reboot.

GBL_SYS_QUEUEThe average number of processes or kernel threads blocked on SYSTM(that is, system resources) during the interval. These resources includedata structures from the LVM, VFS, UFS, JFS, and Disk Quotasubsystems. “SYSTM” is the “catch-all” wait state for blocks on systemresources that are not common enough or long enough to warrant theirown stop state.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SYSTM divided by the interval time.






315

GBL_SYS_WAIT_PCTThe percentage of time processes or kernel threads were blocked onSYSTM (that is, system resources) during the interval. These resourcesinclude data structures from the LVM, VFS, UFS, JFS, and Disk Quotasubsystems. “SYSTM” is the “catch-all” wait state for blocks on systemresources that are not common enough or long enough to warrant theirown stop state.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on SYSTM divided by the accumulated time that allprocesses or kernel threads were alive during the interval.





GBL_SYS_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on SYSTM (that is, system resources) during the interval.These resources include data structures from the LVM, VFS, UFS, JFS,and Disk Quota subsystems. “SYSTM” is the “catch-all” wait state forblocks on system resources that are not common enough or long enoughto warrant their own stop state.

GBL_TERM_IO_QUEUEThe average number of processes or kernel threads blocked on terminalIO (waiting for their terminal IO to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on TERM (that is, terminal IO) divided by theinterval time.



316




GBL_TERM_IO_WAIT_PCTThe percentage of time processes or kernel threads were blocked onterminal IO (waiting for terminal IO to complete) during the interval.

This is calculated as the accumulated time that all processes or kernelthreads spent blocked on TERM (that is, terminal IO) divided by theaccumulated time that all processes or kernel threads were alive duringthe interval.





GBL_TERM_IO_WAIT_TIMEThe accumulated time, in seconds, that all processes or kernel threadswere blocked on terminal IO (waiting for their terminal IO to complete)during the interval.


317

GBL_TT_OVERFLOW_COUNTThe number of new transactions that could not be measured because theMeasurement Processing Daemon's (midaemon) MeasurementPerformance Database is full. If this happens, the default MeasurementPerformance Database size is not large enough to hold all of theregistered transactions on this system. This can be remedied by stoppingand restarting the midaemon process using the -smdvss option to specifya larger Measurement Performance Database size. The currentMeasurement Performance Database size can be checked using themidaemon -sizes option.

LVDETAIL_LABELThe type of entry this volume group or logical volume is associated with -could be a device, partition, file system, logical volume, or volume group.

LVDETAIL_NAMEThe name of the device, partition, file system, logical volume, or volumegroup this volume group or logical volume is associated with.

LV_AVG_READ_SERVICE_TIMEThe average time, in milliseconds, that this logical volume spentprocessing each read request during the interval. For example, a value of5.14 would indicate that read requests during the last interval took onaverage slightly longer than five one-thousandths of a second to completefor this device.

This metric can be used to help determine which logical volumes aretaking more time than usual to process requests.

This metric is reported as “na” for LVM.

On HP-UX 11i and beyond, data is available from VERITAS VolumeManager (VxVM). LVM (Logical Volume Manager) uses the terminology“volume group” to describe a set of related volumes. VERITAS VolumeManager uses the terminology “disk group” to describe a collection of VMdisks. For additional information on VERITAS Volume Manager, seevxintro(1M).

LV_AVG_WRITE_SERVICE_TIMEThe average time, in milliseconds, that this logical volume spentprocessing each write request during the interval. For example, a value of5.14 would indicate that write requests during the last interval took onaverage slightly longer than five one-thousandths of a second to completefor this device.

This metric can be used to help determine which logical volumes aretaking more time than usual to process requests.


318

This metric is reported as “na” for LVM.


LV_CACHE_HITThe number of requests successfully satisfied from the Mirror Write Cache(MWC) during the interval.

The Mirror Write Cache tracks each write of mirrored data to the physicalvolumes and maintains a record of any mirrored writes not yet successfullycompleted at the time of a system crash.

This metric is reported as “na” for VERITAS Volume Manager.


LV_CACHE_MISSThe number of requests that were not satisfied from the Mirror WriteCache (MWC) during the interval.

The MWC is disabled with the lvchange(1M) command (“lvchange -Mn...”), which may increase system performance, but slow down recovery inthe event of a system failure.



LV_CACHE_QUEUEThe number of requests queued to the Mirror Write Cache (MWC) at theend of the interval.

The MWC is only used for volume mirroring and its use degradesperformance, as extra work is required during disk writes to maintain theMirror Write Cache.


319




LV_CACHE_SIZEThe number of entries in this logical volume group's Mirror Write Cache(MWC). The size of this cache is determined by the kernel's logicalvolume code and is not configurable.

The MWC is optional and only used for volume mirroring. The MWCtracks each write of mirrored data to the physical volumes and maintains arecord of any mirrored writes not yet successfully completed at the time ofa system crash.




LV_DEVNOMajor / Minor number of this logical volume.


Disk groups in the VERITAS Volume Manager do not have device files.Therefore, “na” is reported for this metric since it is not applicable.

LV_DIRNAMEThe path name of this logical volume or volume/disk group.


320


For LVM logical volumes, this is the name used as a parameter to thelvdisplay(1M) command. For volume groups, this is the name used as aparameter to the vgdisplay(1M) command.

The entry referred to as the “/dev/vgXX/group” entry shows the internalresources used by the LVM software to manage the logical volumes.

LV_GROUP_NAMEThe name of this volume/disk group associated with a logical volume.

On HP-UX 11i and beyond, data is available from VERITAS VolumeManager (VxVM). LVM (Logical Volume Manager) uses theterminology "volume group" to describe a set of related volumes.VERTIAS Volume Manager uses the terminology "disk group" todescribe a collection of VM disks. For additional information onVERITAS Volume Manager, see vxintro(1M).

LV_INTERVALThe amount of time in the interval.

LV_INTERVAL_CUMThe amount of time over the cumulative collection time, or since the lastconfiguration change.


LV_OPEN_LVThe number of logical volumes currently opened in this volume/disk group.An entry of “na” indicates that there are no logical volumes open in thisvolume group and there are no active disks in this volume group.

The extra entry (referred to as the “/dev/vgXX/group” entry), shows theinternal resources used by the LVM software to manage the logicalvolumes.

On HP-UX 11i and beyond, data is available from VERITAS VolumeManager (VxVM). LVM (Logical Volume Manager) uses the terminology“volume group” to describe a set of related volumes. VERITAS VolumeManager uses the terminology “disk group” to describe a collection of VM


321

disks. For additional information on VERITAS Volume Manager, seevxintro(1M).

LV_READ_BYTE_RATEThe number of physical KBs per second read from this logical volumeduring the interval.

Note that bytes read from the buffer cache are not included in thiscalculation.

LV_READ_BYTE_RATE_CUMThe average number of physical KBs per second read from this logicalvolume over the cumulative collection time, or since the last configurationchange.

Note that bytes read from the buffer cache are not included in thiscalculation.


LV_READ_RATEThe number of physical reads per second for this logical volume during theinterval.

This may not correspond to the physical read rate from a particular diskdrive since a logical volume may be composed of many disk drives or itmay be a subset of a disk drive. An individual physical read from onelogical volume may span multiple individual disk drives.

Since this is a physical read rate, there may not be any correspondence tothe logical read rate since many small reads are satisfied in the buffercache, and large logical read requests must be broken up into physicalread requests.

LV_READ_RATE_CUMThe average number of physical reads per second for this volume over thecumulative collection time, or since the last configuration change.



322

LV_TYPEEither “G” or “V”, indicating either a volume/disk group (“G”) or a logicalvolume (“V”).


LV_WRITE_BYTE_RATEThe number of KBs per second written to this logical volume during theinterval.

LV_WRITE_BYTE_RATE_CUMThe average number of KBs per second written to this logical volume overthe cumulative collection time, or since the last configuration change.


LV_WRITE_RATEThe number of physical writes per second to this logical volume during theinterval.

This may not correspond to the physical write rate to a particular disk drivesince a logical volume may be composed of many disk drives or it may bea subset of a disk drive.

Since this is a physical write rate, there may not be any correspondence tothe logical write rate since many small writes are combined in the buffercache, and many large logical writes must be broken up.

LV_WRITE_RATE_CUMThe average number of physical writes per second to this volume over thecumulative collection time, or since the last configuration change.



323

PRM_BYVG_GROUP_ENTITLEMENTThe PRM Disk entitlement for this PRM Group ID entry as defined in thePRM configuration file. There must be exactly one volume group recordfor every PRM group record. The sum of the disk entitlements must be100 percent for each volume group.

PRM_BYVG_GROUP_UTILA group's current percentage of disk bandwidth relative to other PRMgroups' usage of the same volume group.

PRM_BYVG_INTERVALThe amount of time in the interval.

PRM_BYVG_INTERVAL_CUMThe amount of time over the cumulative collection time.


PRM_BYVG_PRM_GROUPIDThe PRM Group ID. The PRM Group ID is kept in the PRM configurationfile.

PRM_BYVG_PRM_GROUPNAMEThe PRM group name. The PRM group name is kept in the PRMconfiguration file.

PRM_BYVG_REQUESTThe number of KBs (or MBs if specified) the PRM group requested to haveread from or written to the logical volumes in the current volume groupduring the interval.

The PRM_BYVG_* metrics report on the total bytes requested/transferredfor a specified volume group. The byte counts are the total of various IOrequests which result in physical IO activity. These requests may include:

- Raw IO directed to a raw logical

volume

- Delayed Buffer Cache writes


324

- Buffer Cache misses that cause

reads

- Large IO that bypasses buffer

cache

- Virtual Memory Paging Activity

Since the PRM configuration is dynamic, the collection may be restarted.Two intervals are required before the new values are reported. The firstinterval after the collection is restarted displays n/a (not available) for all ofthe counts.

PRM_BYVG_REQUEST_CUMThe number of KBs (or MBs if specified) the PRM group requested beread from or written to the logical volumes in the current volume groupover the cumulative collection time, or since the last configuration change.




volume



reads


cache



PRM_BYVG_REQUEST_QUEUEThe request queue length for the specified volume group.

PRM_BYVG_TRANSFERThe number of KBs (or MBs if specified) the PRM group has read from orwritten to the logical volumes in the current volume group during theinterval.



325


volume



reads


cache



PRM_BYVG_TRANSFER_CUMThe number of KBs (or MBs if specified) the PRM group has read from orwritten to the logical volumes in the current volume group over thecumulative collection time, or since the last configuration change.




volume



reads


cache



PROCSYSCALL_ACTIVE_CUMThe number of different system calls called by this process during the timeit has been enabled for system call profiling.


326

PROCSYSCALL_CALL_COUNTThe number of system calls made to this function by this process duringthe interval.

PROCSYSCALL_CALL_COUNT_CUMThe number of system calls made by this process to this function duringthe time it has been enabled for system call profiling.

PROCSYSCALL_CALL_IDThe ID number of the system call. System calls are sequentiallynumbered starting with one.

PROCSYSCALL_CALL_NAMEThe system call name.

PROCSYSCALL_CALL_RATEThe number of system calls per second made by this process to thisfunction during the last interval.

PROCSYSCALL_CALL_RATE_CUMThe average number of system calls per second made by this process tothis function during the time it has been enabled for system call profiling.

PROCSYSCALL_INTERVALThe amount of time in the interval.

PROCSYSCALL_INTERVAL_CUMThe time, in seconds, system call data has been collected for this process.

PROCSYSCALL_TOTAL_TIMEThe elapsed time, in seconds, this process was in this system call. Thisvalue maybe greater then the interval time since the system call may havebeen started before the interval started.


327

PROCSYSCALL_TOTAL_TIME_CUMThe total elapsed time, in seconds, that this process was in this systemcall. This value maybe greater than the cumulative interval time since thesystem call may have been started before data collection.

PROC_ACTIVE_PROCTHREAD_ACTIVE_THREADThe number of active processes or kernel threads in the interval. Aprocess or kernel thread is active if it is alive and consumes CPU time.

PROC_APP_IDTHREAD_APP_IDThe ID number of the application to which the process or kernel threadbelonged during the interval. Application “other” always has an ID of 1.There can be up to 128 user-defined applications, which are defined in theparm file.

PROC_APP_NAMETHREAD_APP_NAMEThe application name of a process or kernel thread.

Processes or kernel threads are assigned into application groups basedupon rules in the parm file. If a process or kernel thread does not fit anyrules in this file, it is assigned to the application “other.”

The rules include decisions based upon pathname, user ID, priority, andso forth. As these values change during the life of a process or kernelthread, it is re-assigned to another application. This re-evaluation is doneevery measurement interval.

PROC_CACHE_WAIT_PCTTHREAD_CACHE_WAIT_PCTThe percentage of time the process or kernel thread was blocked onCACHE (waiting for the file system buffer cache to be updated) during theinterval. Processes or kernel threads doing raw IO to a disk are notincluded in this measurement. Processes and kernel threads doingbuffered IO to disks normally spend more time blocked on CACHE and IOthan on DISK.

On a threaded operating system, such as HP-UX 11.0 and beyond,process wait time is calculated by summing the wait times of its kernelthreads. Alive kernel threads and kernel threads that have died during theinterval are included in the summation.

A percentage of time spent in a wait state is calculated as the time akernel thread (or all kernel threads of a process) spent waiting in this state,


328

divided by the alive time of the kernel thread (or all kernel threads of theprocess) during the interval.

If this metric is reported for a kernel thread, the percentage value is for thatsingle kernel thread. If this metric is reported for a process, thepercentage value is calculated with the sum of the wait and alive times ofall of its kernel threads.

For example, if a process has 2 kernel threads, one sleeping for the entireinterval and one waiting on terminal input for the interval, the process waitpercent values will be 50% on Sleep and 50% on Terminal. The kernelthread wait values will be 100% on Sleep for the first kernel thread and100% on Terminal for the second kernel thread.

For another example, consider the same process as above, with 2 kernelthreads, one of which was created half-way through the interval, and whichthen slept for the remainder of the interval. The other kernel thread waswaiting for terminal input for half the interval, then used the CPU activelyfor the remainder of the interval. The process wait percent values will be33% on Sleep and 33% on Terminal (each one third of the total alive time).The kernel thread wait values will be 100% on Sleep for the first kernelthread and 50% on Terminal for the second kernel thread.

PROC_CACHE_WAIT_PCT_CUMTHREAD_CACHE_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon CACHE (waiting for the file system buffer cache to be updated) overthe cumulative collection time. Processes or kernel threads doing raw IOto a disk are not included in this measurement.



A percentage of time spent in a wait state is calculated as the time akernel thread (or all kernel threads of a process) spent waiting in this state,divided by the alive time of the kernel thread (or all kernel threads of theprocess) during the interval.




329


PROC_CACHE_WAIT_TIMETHREAD_CACHE_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onCACHE (waiting for the file system buffer cache to be updated) during theinterval. Processes or kernel threads doing raw IO to a disk are notincluded in this measurement.

On a threaded operating system, such as HP-UX 11.0 and beyond,process wait time is calculated by summing the wait times of its kernelthreads. If this metric is reported for a kernel thread, the value is the waittime of that single kernel thread. If this metric is reported for a process,the value is the sum of the wait times of all of its kernel threads. Alivekernel threads and kernel threads that have died during the interval areincluded in the summation. For multi-threaded processes, the wait timescan exceed the length of the measurement interval.

PROC_CACHE_WAIT_TIME_CUMTHREAD_CACHE_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onCACHE (waiting for the file system buffer cache to be updated) over thecumulative collection time. Processes or kernel threads doing raw IO to adisk are not included in this measurement.




330

PROC_CDFS_WAIT_PCTTHREAD_CDFS_WAIT_PCTThe percentage of time the process or kernel thread was blocked onCDFS (waiting for its Compact Disk file system IO to complete) during theinterval.






PROC_CDFS_WAIT_PCT_CUMTHREAD_CDFS_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon CDFS (waiting for its Compact Disk file system IO to complete) over thecumulative collection time.





331




PROC_CDFS_WAIT_TIMETHREAD_CDFS_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onCDFS (waiting in the CD-ROM driver for Compact Disc file system IO tocomplete) during the interval.


PROC_CDFS_WAIT_TIME_CUMTHREAD_CDFS_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onCDFS (waiting in the CD-ROM driver for Compact Disc file system IO tocomplete) over the cumulative collection time.


On a threaded operating system, such as HP-UX 11.0 and beyond,process wait time is calculated by summing the wait times of its kernelthreads. If this metric is reported for a kernel thread, the value is the waittime of that single kernel thread. If this metric is reported for a process,the value is the sum of the wait times of all of its kernel threads. Alivekernel threads and kernel threads that have died during the interval are


332

included in the summation. For multi-threaded processes, the wait timescan exceed the length of the measurement interval.

PROC_CLOSETHREAD_CLOSEThe number of file closes made by the process or kernel thread during theinterval. This corresponds to the number of close(2) system calls.

This metric is specific to a process. If this metric is reported for a kernelthread, the value for its associated process is given.

PROC_CLOSE_CUMTHREAD_CLOSE_CUMThe number of file closes made by the process or kernel thread over thecumulative collection time. This corresponds to the number of close(2)system calls.



PROC_CPU_CSWITCH_TIMETHREAD_CPU_CSWITCH_TIMEThe time, in seconds, that the process or kernel thread spent in contextswitching during the interval.

On a threaded operating system, such as HP-UX 11.0 and beyond,process usage of a resource is calculated by summing the usage of thatresource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_CPU_CSWITCH_TIME_CUMTHREAD_CPU_CSWITCH_TIME_CUMThe time, in seconds, that the selected process or kernel thread spent incontext switching over the cumulative collection time.



333


PROC_CPU_CSWITCH_UTILTHREAD_CPU_CSWITCH_UTILThe percentage of time spent in context switching the current process orkernel thread during the interval.

On a threaded operating system, such as HP-UX 11.0 and beyond,process usage of a resource is calculated by summing the usage of thatresource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation. On multi-processorsystems, processes which have component kernel threads executingsimultaneously on different processors could have resource utilizationsums over 100%.

PROC_CPU_CSWITCH_UTIL_CUMTHREAD_CPU_CSWITCH_UTIL_CUMThe average percentage of time spent in context switching the process orkernel thread over the cumulative collection time.



PROC_CPU_INTERRUPT_TIMETHREAD_CPU_INTERRUPT_TIMEThe time, in seconds, that the process or kernel thread spent processinginterrupts during the interval.


334


PROC_CPU_INTERRUPT_TIME_CUMTHREAD_CPU_INTERRUPT_TIME_CUMThe time, in seconds, that the process or kernel thread spent processinginterrupts over the cumulative collection time.



PROC_CPU_INTERRUPT_UTILTHREAD_CPU_INTERRUPT_UTILThe percentage of time that this process or kernel thread was in interruptmode during the last interval. Interrupt mode means that interrupts werebeing handled while the process or kernel thread was loaded and runningon the CPU. The interrupts may have been generated by any process, notjust the running process, but they were handled while the process orkernel thread was running and may have had an impact on theperformance of this process or kernel thread.



335

PROC_CPU_INTERRUPT_UTIL_CUMTHREAD_CPU_INTERRUPT_UTIL_CUMThe average percentage of time that this process or kernel thread was ininterrupt mode over the cumulative collection time.



PROC_CPU_LAST_USEDTHREAD_CPU_LAST_USEDThe ID number of the processor that last ran the process or kernel thread.For uni-processor systems, this value is always zero.

On a threaded operating system, such as HP-UX 11.0 and beyond, thismetric represents a kernel thread characteristic. If this metric is reportedfor a process, the value for its last executing kernel thread is given. Forexample, if a process has multiple kernel threads and kernel thread one isthe last to execute during the interval, the metric value for kernel threadone is assigned to the process.

PROC_CPU_NICE_TIMETHREAD_CPU_NICE_TIMEThe time, in seconds, that this niced process or kernel thread was usingthe CPU in user mode during the interval. The NICE metrics includepositive nice value CPU time only. Negative nice value CPU is broken outinto NNICE (negative nice) metrics. Positive nice values range from 20 to39. Negative nice values range from 0 to 19.



336

PROC_CPU_NICE_TIME_CUMTHREAD_CPU_NICE_TIME_CUMThe time, in seconds, that this niced process or kernel thread was in usermode over the cumulative collection time. The NICE metrics includepositive nice value CPU time only. Negative nice value CPU is broken outinto NNICE (negative nice) metrics. Positive nice values range from 20 to39. Negative nice values range from 0 to 19.



PROC_CPU_NICE_UTILTHREAD_CPU_NICE_UTILThe percentage of time that this niced process or kernel thread was inuser mode during the interval. The NICE metrics include positive nicevalue CPU time only. Negative nice value CPU is broken out into NNICE(negative nice) metrics. Positive nice values range from 20 to 39.Negative nice values range from 0 to 19.


PROC_CPU_NICE_UTIL_CUMTHREAD_CPU_NICE_UTIL_CUMThe average percentage of time that this niced process or kernel threadwas in user mode over the cumulative collection time. The NICE metricsinclude positive nice value CPU time only. Negative nice value CPU isbroken out into NNICE (negative nice) metrics. Positive nice values rangefrom 20 to 39. Negative nice values range from 0 to 19.



337



PROC_CPU_NNICE_TIMETHREAD_CPU_NNICE_TIMEThe time, in seconds, that this negatively niced process or kernel threadwas using the CPU in user mode during the interval.


PROC_CPU_NNICE_TIME_CUMTHREAD_CPU_NNICE_TIME_CUMThe time, in seconds, that this negatively niced process or kernel threadwas in user mode over the cumulative collection time.



PROC_CPU_NNICE_UTILTHREAD_CPU_NNICE_UTILThe percentage of time that this negatively niced process or kernel threadwas in user mode during the interval.


338


PROC_CPU_NNICE_UTIL_CUMTHREAD_CPU_NNICE_UTIL_CUMThe average percentage of time that this negatively niced process orkernel thread was in user mode over the cumulative collection time.



PROC_CPU_NORMAL_TIMETHREAD_CPU_NORMAL_TIMEThe time, in seconds, that the selected process or kernel thread was inuser mode at normal priority during the interval. Normal priority user modeCPU excludes CPU used at real-time and nice priorities.



339

PROC_CPU_NORMAL_TIME_CUMTHREAD_CPU_NORMAL_TIME_CUMThe time, in seconds, that the selected process or kernel thread was inuser mode at normal priority over the cumulative collection time. Normalpriority user mode CPU excludes CPU used at real-time and nice priorities.



PROC_CPU_NORMAL_UTILTHREAD_CPU_NORMAL_UTILThe percentage of time that this process or kernel thread was in usermode at a normal priority during the interval. “At a normal priority” meansthe neither rtprio or nice had been used to alter the priority of the processor kernel thread during the interval. Normal priority user mode CPUexcludes CPU used at real-time and nice priorities.


PROC_CPU_NORMAL_UTIL_CUMTHREAD_CPU_NORMAL_UTIL_CUMThe average percentage of time a process or kernel thread was in usermode at normal priority over the cumulative collection time. Normal priorityuser mode CPU excludes CPU used at real-time and nice priorities.


On a threaded operating system, such as HP-UX 11.0 and beyond,process usage of a resource is calculated by summing the usage of that


340

resource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation. On multi-processorsystems, processes which have component kernel threads executingsimultaneously on different processors could have resource utilizationsums over 100%.

PROC_CPU_REALTIME_TIMETHREAD_CPU_REALTIME_TIMEThe time, in seconds, that the selected process or kernel thread was inuser mode at a realtime priority during the interval.


PROC_CPU_REALTIME_TIME_CUMTHREAD_CPU_REALTIME_TIME_CUMThe time, in seconds, that the selected process or kernel thread was inuser mode at a realtime priority over the cumulative collection time.



PROC_CPU_REALTIME_UTILTHREAD_CPU_REALTIME_UTILThe percentage of time that this process or kernel thread was at a realtimepriority during the interval. The realtime CPU is separated out to allowusers to see the effect of using the realtime facilities to alter priority.

On a threaded operating system, such as HP-UX 11.0 and beyond,process usage of a resource is calculated by summing the usage of thatresource by its kernel threads. If this metric is reported for a kernel thread,


341

the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation. On multi-processorsystems, processes which have component kernel threads executingsimultaneously on different processors could have resource utilizationsums over 100%.

PROC_CPU_REALTIME_UTIL_CUMTHREAD_CPU_REALTIME_UTIL_CUMThe percentage of time that the CPU was in user mode executing thecurrent process or kernel thread at a realtime priority over the cumulativecollection time.



PROC_CPU_SWITCHESTHREAD_CPU_SWITCHESThe number of times the process or kernel thread was switched to anotherprocessor during the interval. For uni-processor systems, this value isalways zero.


PROC_CPU_SWITCHES_CUMTHREAD_CPU_SWITCHES_CUMThe number of times the process or kernel thread was switched to anotherprocessor over the cumulative collection time. For uni-processor systems,this value is always zero.


342



PROC_CPU_SYSCALL_TIMETHREAD_CPU_SYSCALL_TIMEThe time, in seconds, that this process or kernel thread spent executingsystem calls in system mode, excluding interrupt or context processing,during the interval.


PROC_CPU_SYSCALL_TIME_CUMTHREAD_CPU_SYSCALL_TIME_CUMThe time, in seconds, that this process or kernel thread spent executingsystem calls in system mode, excluding interrupt or context processing,over the cumulative collection time.




343

PROC_CPU_SYSCALL_UTILTHREAD_CPU_SYSCALL_UTILThe percentage of the total CPU time this process or kernel thread spentexecuting system calls in system mode during the interval.


PROC_CPU_SYSCALL_UTIL_CUMTHREAD_CPU_SYSCALL_UTIL_CUMThe average percentage of the total CPU time this process or kernelthread spent executing system calls in system mode over the cumulativecollection time.



PROC_CPU_SYS_MODE_TIMETHREAD_CPU_SYS_MODE_TIMEThe CPU time in system mode in the context of the process or kernelthread during the interval.



344

PROC_CPU_SYS_MODE_TIME_CUMTHREAD_CPU_SYS_MODE_TIME_CUMThe CPU time in system mode in the context of the process or kernelthread over the cumulative collection time.



PROC_CPU_SYS_MODE_UTILTHREAD_CPU_SYS_MODE_UTILThe percentage of time that the CPU was in system mode in the context ofthe process or kernel thread during the interval.

Unlike the global and application CPU metrics, process CPU is notaveraged over the number of processors on systems with multiple CPUs.Single-threaded processes can use only one CPU at a time and neverexceed 100% CPU utilization.

High system mode CPU utilizations are normal for IO intensive programs.Abnormally high system CPU utilization can indicate that a hardwareproblem is causing a high interrupt rate. It can also indicate programs thatare not using system calls efficiently.

A classic “hung shell” shows up with very high system mode CPU becauseit gets stuck in a loop doing terminal reads (a system call) to a device thatnever responds.


On multi-processor systems, processes which have component kernelthreads executing simultaneously on different processors could haveresource utilization sums over 100%.


345

PROC_CPU_SYS_MODE_UTIL_CUMTHREAD_CPU_SYS_MODE_UTIL_CUMThe average percentage of time that the CPU was in system mode in thecontext of the process or kernel thread over the cumulative collection time.





PROC_CPU_TOTAL_TIMETHREAD_CPU_TOTAL_TIMEThe total CPU time, in seconds, consumed by a process or kernel threadduring the interval.

Total CPU time is the sum of the CPU time components for a process orkernel thread, including system, user, context switch, interrupt processing,realtime, and nice utilization values.


PROC_CPU_TOTAL_TIME_CUMTHREAD_CPU_TOTAL_TIME_CUMThe total CPU time consumed by a process or kernel thread over thecumulative collection time. CPU time is in seconds unless otherwisespecified.



346



PROC_CPU_TOTAL_TIME_CUM =

PROC_CPU_SYS_MODE_TIME_CUM +

PROC_CPU_USER_MODE_TIME_CUM


PROC_CPU_TOTAL_UTILTHREAD_CPU_TOTAL_UTILThe total CPU time consumed by a process or kernel thread as apercentage of the total CPU time available during the interval.


Total CPU utilization is the sum of the CPU utilization components for aprocess or kernel thread, including system, user, context switch interruptsprocessing, realtime, and nice utilization values.



PROC_CPU_TOTAL_UTIL_CUMTHREAD_CPU_TOTAL_UTIL_CUMThe total CPU time consumed by a process or kernel thread as apercentage of the total CPU time available over the cumulative collectiontime.



347



Total CPU utilization is the sum of the CPU utilization components for aprocess or kernel thread, including system, user, context switch, interruptprocessing, realtime, and nice utilization values.



PROC_CPU_TRAP_COUNTTHREAD_CPU_TRAP_COUNTThe number of times the CPU was in trap handler code for this process orkernel thread during the interval.

All exceptions (including faults) cause traps. These include pfaults(protection faults), vfaults (virtual faults), time slice expiration(rescheduling), zero divide, illegal or privileged instructions, single-stepping, breakpoints, and so on. The kernel trap handler code will switchtrap counters for vfaults and pfaults to fault counters when appropriate. Assuch, the trap count excludes vfaults and pfaults.


PROC_CPU_TRAP_COUNT_CUMTHREAD_CPU_TRAP_COUNT_CUMThe number of times the CPU was in trap handler code for this process orkernel thread over the cumulative collection time.



348



PROC_CPU_USER_MODE_TIMETHREAD_CPU_USER_MODE_TIMEThe time, in seconds, the process or kernel thread was using the CPU inuser mode during the interval.



PROC_CPU_USER_MODE_TIME_CUMTHREAD_CPU_USER_MODE_TIME_CUMThe time, in seconds, the process or kernel thread was using the CPU inuser mode over the cumulative collection time.





349

PROC_CPU_USER_MODE_UTILTHREAD_CPU_USER_MODE_UTILThe percentage of time the process or kernel thread was using the CPU inuser mode during the interval.

Unlike the global and application CPU metrics, process CPU is notaveraged over the number of processors on systems with multiple CPUs.Single-threaded processes can use only one CPU at a time and neverexceed 100% CPU utilization. User CPU is the time spent in user mode ata normal priority, at real-time priority, and at a nice priority.



PROC_CPU_USER_MODE_UTIL_CUMTHREAD_CPU_USER_MODE_UTIL_CUMThe average percentage of time the process or kernel thread was usingthe CPU in user mode over the cumulative collection time.







350

PROC_DISK_FS_READTHREAD_DISK_FS_READNumber of file system physical disk reads made by a process or kernelthread during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.


“Disk” in this instance refers to any locally attached physical disk drives(that is, “spindles”) that may hold file systems and/or swap. NFS mounteddisks are not included in this list.


PROC_DISK_FS_READ_CUMTHREAD_DISK_FS_READ_CUMNumber of file system physical disk reads made by a process or kernelthread over the cumulative collection time. Only local disks are counted inthis measurement. NFS devices are excluded.






351

PROC_DISK_FS_READ_RATETHREAD_DISK_FS_READ_RATEThe number of file system physical disk reads made by a process orkernel thread during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.




PROC_DISK_FS_WRITETHREAD_DISK_FS_WRITENumber of file system physical disk writes made by a process or kernelthread during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.






352

PROC_DISK_FS_WRITE_CUMTHREAD_DISK_FS_WRITE_CUMNumber of file system physical disk writes made by a process or kernelthread over the cumulative collection time. Only local disks are counted inthis measurement. NFS devices are excluded.






PROC_DISK_FS_WRITE_RATETHREAD_DISK_FS_WRITE_RATEThe number of file system physical disk writes made by a process orkernel thread during the interval. Only local disks are counted in thismeasurement. NFS devices are excluded.





353

PROC_DISK_LOGL_IOTHREAD_DISK_LOGL_IOThe number of logical IOs made by (or for) a process or kernel threadduring the interval. NFS mounted disks are not included in this list.




PROC_DISK_LOGL_IO_CUMTHREAD_DISK_LOGL_IO_CUMThe number of logical IOs made by (or for) a process or kernel thread overthe cumulative collection time. NFS mounted disks are not included in thislist.






354

PROC_DISK_LOGL_IO_RATETHREAD_DISK_LOGL_IO_RATEThe number of logical IOs per second made by (or for) a process or kernelthread during the interval. NFS mounted disks are not included in this list.




For processes which run for less than the measurement interval, thismetric is normalized over the measurement interval. For example, aprocess ran for 1 second and did 50 IOs during its life. If themeasurement interval is 5 seconds, it is reported as having done 10 IOsper second. If the measurement interval is 60 seconds, it is reported ashaving done 50/60 or 0.83 IOs per second.



PROC_DISK_LOGL_IO_RATE_CUMTHREAD_DISK_LOGL_IO_RATE_CUMThe average number of logical IOs per second made by (or for) a processor kernel thread over the cumulative collection time. Only local disks arecounted in this measurement. NFS devices are excluded.


Logical disk IOs are measured by counting the read and write systemcalls that are directed to disk devices. Also counted are read and write


355

system calls made indirectly through other system calls, including readv,recvfrom, recv, recvmsg, ipcrecvcn, recfrom, writev, send, sento,sendmsg, and ipcsend.




PROC_DISK_LOGL_READTHREAD_DISK_LOGL_READThe number of disk logical reads made by a process or kernel threadduring the interval. Calls destined for NFS mounted files are not counted.



PROC_DISK_LOGL_READ_CUMTHREAD_DISK_LOGL_READ_CUMThe number of disk logical reads made by a process or kernel thread overthe cumulative collection time. Calls destined for NFS mounted files arenot counted.



356




PROC_DISK_LOGL_READ_RATETHREAD_DISK_LOGL_READ_RATEThe number of logical reads per second made by (or for) a process orkernel thread during the interval. Calls destined for NFS mounted files arenot counted.




PROC_DISK_LOGL_WRITETHREAD_DISK_LOGL_WRITENumber of disk logical writes made by a process or kernel thread duringthe interval. Calls destined for NFS mounted files are not counted.




357


PROC_DISK_LOGL_WRITE_CUMTHREAD_DISK_LOGL_WRITE_CUMNumber of disk logical writes made by a process or kernel thread over thecumulative collection time. Calls destined for NFS mounted files are notcounted.




PROC_DISK_LOGL_WRITE_RATETHREAD_DISK_LOGL_WRITE_RATEThe number of logical writes per second made by (or for) a process orkernel thread during the interval. NFS mounted disks are not included inthis list.





358

PROC_DISK_PHYS_IO_RATETHREAD_DISK_PHYS_IO_RATEThe average number of physical disk IOs per second made by the processor kernel thread during the interval.

For processes which run for less than the measurement interval, thismetric is normalized over the measurement interval. For example, aprocess ran for 1 second and did 50 IOs during its life. If themeasurement interval is 5 seconds, it is reported as having done 10 IOsper second. If the measurement interval is 60 seconds, it is reported ashaving done 50/60 or 0.83 IOs per second.




PROC_DISK_PHYS_IO_RATE_CUMTHREAD_DISK_PHYS_IO_RATE_CUMThe number of physical disk IOs per second made by the selected processor kernel thread over the cumulative collection time.






359

PROC_DISK_PHYS_READTHREAD_DISK_PHYS_READThe number of physical reads made by (or for) a process or kernel threadduring the last interval.

“Disk” refers to a physical drive (that is, “spindle”), not a partition on adrive (unless the partition occupies the entire physical disk). NFSmounted disks are not included in this list.


PROC_DISK_PHYS_READ_CUMTHREAD_DISK_PHYS_READ_CUMThe number of physical reads made by (or for) a process or kernel threadover the cumulative collection time.




PROC_DISK_PHYS_READ_RATETHREAD_DISK_PHYS_READ_RATEThe number of physical reads per second made by (or for) a process orkernel thread during the interval.




360

resource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_DISK_PHYS_WRITETHREAD_DISK_PHYS_WRITEThe number of physical writes made by (or for) a process or kernel threadduring the last interval.




PROC_DISK_PHYS_WRITE_CUMTHREAD_DISK_PHYS_WRITE_CUMThe number of physical writes made by (or for) a process or kernel threadover the cumulative collection time.






361

PROC_DISK_PHYS_WRITE_RATETHREAD_DISK_PHYS_WRITE_RATEThe number of physical writes per second made by (or for) a process orkernel thread during the interval.



PROC_DISK_RAW_READTHREAD_DISK_RAW_READNumber of raw reads made for a process or kernel thread during theinterval.




PROC_DISK_RAW_READ_CUMTHREAD_DISK_RAW_READ_CUMNumber of raw reads made for a process or kernel thread over thecumulative collection time.




362



PROC_DISK_RAW_READ_RATETHREAD_DISK_RAW_READ_RATERate of raw reads made for a process or kernel thread during the interval.




PROC_DISK_RAW_WRITETHREAD_DISK_RAW_WRITENumber of raw writes made for a process or kernel thread during theinterval.





363

PROC_DISK_RAW_WRITE_CUMTHREAD_DISK_RAW_WRITE_CUMNumber of raw writes made for a process or kernel thread over thecumulative collection time.





PROC_DISK_RAW_WRITE_RATETHREAD_DISK_RAW_WRITE_RATERate of raw writes made for a process or kernel thread during the interval.




PROC_DISK_REM_LOGL_READTHREAD_DISK_REM_LOGL_READThe number of remote logical reads made by a process or kernel threadduring the last interval.

Remote logical IOs include all IO requests generated on a local client to aremotely mounted file system or disk. If the logical request is satisfied on


364

the local client (that is, the data is in a local memory buffer), a physicalrequest is not generated. Otherwise, a physical IO request is made to theremote machine to read/write the data. Note that, in either case, a logicalIO request is made.


PROC_DISK_REM_LOGL_READ_CUMTHREAD_DISK_REM_LOGL_READ_CUMThe number of remote logical reads made by a process or kernel threadover the cumulative collection time.




PROC_DISK_REM_LOGL_READ_RATETHREAD_DISK_REM_LOGL_READ_RATEThe number of remote logical reads per second made by (or for) a processor kernel thread during the interval.




365

the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_DISK_REM_LOGL_WRITETHREAD_DISK_REM_LOGL_WRITENumber of remote logical writes made by a process or kernel threadduring the interval.



PROC_DISK_REM_LOGL_WRITE_CUMTHREAD_DISK_REM_LOGL_WRITE_CUMNumber of remote logical writes made by a process or kernel thread overthe cumulative collection time.





366

PROC_DISK_REM_LOGL_WRITE_RATETHREAD_DISK_REM_LOGL_WRITE_RATEThe number of remote logical writes per second made by (or for) aprocess or kernel thread during the interval.



PROC_DISK_REM_PHYS_READTHREAD_DISK_REM_PHYS_READThe number of remote physical reads made by (or for) a process or kernelthread during the interval.



PROC_DISK_REM_PHYS_READ_CUMTHREAD_DISK_REM_PHYS_READ_CUMThe number of remote physical reads made by (or for) a process or kernelthread over the cumulative collection time.




367


PROC_DISK_REM_PHYS_READ_RATETHREAD_DISK_REM_PHYS_READ_RATEThe number of remote physical reads per second made by (or for) aprocess or kernel thread during the interval.



PROC_DISK_REM_PHYS_WRITETHREAD_DISK_REM_PHYS_WRITEThe number of physical writes made by (or for) a process or kernel threadduring the interval.



PROC_DISK_REM_PHYS_WRITE_CUMTHREAD_DISK_REM_PHYS_WRITE_CUMThe number of physical writes made by (or for) a process or kernel threadover the cumulative collection time.


368




PROC_DISK_REM_PHYS_WRITE_RATETHREAD_DISK_REM_PHYS_WRITE_RATEThe number of physical writes per second made by (or for) a process orkernel thread during the interval.



PROC_DISK_SUBSYSTEM_WAIT_PCTTHREAD_DISK_SUBSYSTEM_WAIT_PCTThe percentage of time the process or kernel thread was blocked on thedisk subsystem (waiting for its file system IOs to complete) during theinterval. This includes time spent waiting in the DISK, INODE, CACHE,and CDFS wait states. It does not include processes doing raw IO to diskdevices.




369




PROC_DISK_SUBSYSTEM_WAIT_PCT_CUMTHREAD_DISK_SUBSYSTEM_WAIT_PCT_CUMThe percentage of time the process or kernel thread was blocked on thedisk subsystem (waiting for its file system IOs to complete) over thecumulative collection time. This includes time spent waiting in the DISK,INODE, CACHE, and CDFS wait states. It does not include processesdoing raw IO to disk devices.






For another example, consider the same process as above, with 2 kernelthreads, one of which was created half-way through the interval, and whichthen slept for the remainder of the interval. The other kernel thread was


370

waiting for terminal input for half the interval, then used the CPU activelyfor the remainder of the interval. The process wait percent values will be33% on Sleep and 33% on Terminal (each one third of the total alive time).The kernel thread wait values will be 100% on Sleep for the first kernelthread and 50% on Terminal for the second kernel thread.

PROC_DISK_SUBSYSTEM_WAIT_TIMETHREAD_DISK_SUBSYSTEM_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked on thedisk subsystem (waiting for its file system IOs to complete) during theinterval. This includes time spent waiting in the DISK, INODE, CACHE,and CDFS wait states. It does not include processes doing raw IO to diskdevices.


PROC_DISK_SUBSYSTEM_WAIT_TIME_CUMTHREAD_DISK_SUBSYSTEM_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked on thedisk subsystem (waiting for its file system IOs to complete) over thecumulative collection time. This includes time spent waiting in the DISK,INODE, CACHE, and CDFS wait states. It does not include processesdoing raw IO to disk devices.



PROC_DISK_SYSTEM_IOTHREAD_DISK_SYSTEM_IONumber of file system management physical disk IOs made for a processor kernel thread during the interval.


371





PROC_DISK_SYSTEM_IO_RATETHREAD_DISK_SYSTEM_IO_RATEThe number of file system management physical disk IOs per secondmade for a process or kernel thread during the interval.





PROC_DISK_SYSTEM_READTHREAD_DISK_SYSTEM_READNumber of file system management physical disk reads made for aprocess or kernel thread during the interval.



372



PROC_DISK_SYSTEM_READ_CUMTHREAD_DISK_SYSTEM_READ_CUMNumber of file system management physical disk reads made for aprocess or kernel thread over the cumulative collection time.





PROC_DISK_SYSTEM_WRITETHREAD_DISK_SYSTEM_WRITENumber of file system management physical disk writes made for aprocess or kernel thread during the interval.





373


PROC_DISK_SYSTEM_WRITE_CUMTHREAD_DISK_SYSTEM_WRITE_CUMNumber of file system management physical disk writes made for aprocess or kernel thread over the cumulative collection time.






PROC_DISK_VM_IOTHREAD_DISK_VM_IOThe number of virtual memory IOs made for a process or kernel threadduring the interval.





374

the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_DISK_VM_IO_RATETHREAD_DISK_VM_IO_RATEThe number of virtual memory IOs per second made for a process orkernel thread during the interval.




PROC_DISK_VM_READTHREAD_DISK_VM_READNumber of virtual memory reads made for a process or kernel threadduring the interval.



PROC_DISK_VM_READ_CUMTHREAD_DISK_VM_READ_CUMNumber of virtual memory reads made for a process or kernel thread overthe cumulative collection time.



375




PROC_DISK_VM_WRITETHREAD_DISK_VM_WRITENumber of virtual memory writes made for a process or kernel threadduring the interval.




PROC_DISK_VM_WRITE_CUMTHREAD_DISK_VM_WRITE_CUMNumber of virtual memory writes made for a process or kernel thread overthe cumulative collection time.





376


PROC_DISK_WAIT_PCTTHREAD_DISK_WAIT_PCTThe percentage of time the process or kernel thread was blocked on DISK(waiting in the disk drivers for file system disk IO to complete) during theinterval. The time spent waiting in the disk drivers is usually very small.Most of the time, processes doing file system IO are waiting on IO orCACHE. Processes waiting for character (raw) IO to a disk device areusually waiting on IO.






PROC_DISK_WAIT_PCT_CUMTHREAD_DISK_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon DISK (waiting in the disk drivers for file system disk IO to complete)over the cumulative collection time.


377







PROC_DISK_WAIT_TIMETHREAD_DISK_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onDISK (waiting in a disk driver for its disk IO to complete) during theinterval.



378

PROC_DISK_WAIT_TIME_CUMTHREAD_DISK_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onDISK (waiting in a disk driver for its disk IO to complete) over thecumulative collection time.



PROC_DISPATCHTHREAD_DISPATCHThe number of times the process or kernel thread was made the executingprocess on the CPU over the interval. This includes dispatchesassociated with a context switch because some other process or kernelthread had the CPU, as well as those dispatches caused by the process orkernel thread stopping, then resuming, with no other process or kernelthread running in the meantime.


PROC_DISPATCH_CUMTHREAD_DISPATCH_CUMThe number of times the process or kernel thread was made the executingprocess on the CPU over the cumulative collection time. This includesdispatches associated with a context switch because some other processor kernel thread had the CPU, as well as those dispatches caused by theprocess or kernel thread stopping, then resuming, with no other process orkernel thread running in the meantime.



379


PROC_EUIDTHREAD_EUIDThe Effective User ID of a process.


PROC_FILE_COUNTThe number of times this file is opened currently. Terminal devices areoften opened more than once by several different processes.

PROC_FILE_MODEA text string summarizing the type of open mode:

rd/wr Opened for input & output

read Opened for input only

write Opened for output only

PROC_FILE_NAMEThe path name or identifying information about the open file descriptor. Ifthe path name string exceeds 40 characters in length, the beginning andthe end of the path is shown and the middle of the name is replaced by“...”.

An attempt is made to obtain the file path name by either searching thecurrent cylinder group to find directory entries that point to the currentlyopened inode, or by searching the kernel name cache. Since looking upfile path names would require high disk overhead, some names may notbe resolved. If the path name can not be resolved, a string is returnedindicating the type and inode number of the file.

For HP-UX 11.0 releases, the path name information of a file descriptormay correspond to streams device files, such as /dev/tcp and /dev/udp,which are not explicitly opened by the process. These files are opened bynetworking functions called by the process to access remote systems.

For the string format including an inode number, you may use thencheck(1M) program to display the file path name relative to the mountpoint. Sometimes files may be deleted before they are closed. In these


380

cases, the process file table may still have the inode even though the file isnot actually present and as a result, ncheck will fail.

In the following example, note that the file system name has been includedto avoid the overhead of ncheck searching all of the file systems for theinode number.

If the following file information was displayed:

Note that the following examples would all appear on one line.

<reg,vxfs,/var,/dev/vg00/lvol8,inode:702>

and then from that display, the following ncheck command was entered:

ncheck -i 702 -F vxfs /dev/vg00/lvol8

An output like the following would be generated:

/dev/vg00/lvol8:

702 /adm/cron/log

Since in this example /var is mounted on lvol8 of vg00, the full path namewould be /var/adm/cron/log.

The string shown representing inode information when the path is notavailable is as follows:

<type,domain,filesys,volume,inode:n>

where:

The file type can be one of:

blk - Block device

chr - Character device

dir - Directory file

fifo - FIFO

lnk - Soft file link

reg - Regular file

sock - Socket

emptydir - Mknod created

directory - no

files

ukn - Unknown vtype

The file domain can be one of:

ufs - UNIX file system

nfs - NFS

vxfs - Veritas file system

cdfs - CDROM file system

nfs_spec - NFS special device

file

nfs_bdev - NFS device file -


381

block mode access

nfs_fifo - FIFO file access

over NFS

dev_vn - Generic vnode -

temp type used by

kernel

dummy - OSF's file on file

mount file system

pipe - Pipe

ukn - Unknown vfs type

The filesys is the file system mount point.

The volume field indicates the logical volume, if applicable.

For HP-UX 10.30 and earlier releases, if the file descriptor represents anopen socket, the output format will contain the domain and protocol,followed by the IP address.

For HP-UX 11.0 and later, if the file descriptor represents a unix addressfamily socket which is used for IPC on the local host, its path name will beresolved if possible. For example:

unix /tmp/.AgentSockets/A

If the local socket pathname cannot be resolved, the socket address willbe shown, for example:

unix -> 0x0339a200

For HP-UX 11.0 and later, if the file descriptor is a socket for internetworkcommunications (for example, udp or tcp), the socket address, domain,and protocol will be displayed respectively. The bound IP address andport number will also be displayed when available. For example:

<socket: 0x03189800,inet,tcp,INADDR_ANY:2121>

<socket: 0x031bd400,inet,udp,15.8.157.15:123>

PROC_FILE_NUMBERThe file number of the current open file.

PROC_FILE_OFFSETThe decimal value of the next access position of the current file at the endof the interval. If the open file is a tty, this is the total number of bytes sentand received since the file was first opened.

PROC_FILE_OPENNumber of files the current process has remaining open as of the end ofthe interval.


382

PROC_FILE_TYPEA text string describing the type of the current file. This is one of:

block Block special device

chr Character device

dir Directory

fifo FIFO

file Simple file

link Symbolic File link

network Network channel device

other An unknown file type

pipe Named pipe (FIFO)

reg Regular file

socket Socket

streams Streams

PROC_FORCED_CSWITCHTHREAD_FORCED_CSWITCHThe number of times that the process or kernel thread was preempted byan external event and another process or kernel thread was allowed toexecute during the interval.

Examples of reasons for a forced switch include expiration of a time sliceor returning from a system call with a higher priority process or kernelthread ready to run.


PROC_FORCED_CSWITCH_CUMTHREAD_FORCED_CSWITCH_CUMThe number of times the process or kernel thread was preempted by anexternal event and another process or kernel thread was allowed toexecute over the cumulative collection time.

Examples of reasons for a forced switch include expiration of a time sliceor returning from a system call with a higher priority process or kernelthread ready to run.



383


PROC_FORKTHREAD_FORKThe total number of fork and vfork system calls executed by this processduring the interval.


PROC_FORK_CUMTHREAD_FORK_CUMThe number of fork or vfork system calls made by a process over thecumulative collection time.



PROC_GRAPHICS_WAIT_PCTTHREAD_GRAPHICS_WAIT_PCTThe percentage of time the process or kernel thread was blocked ongraphics (waiting for graphics operations to complete) during the interval.




For example, if a process has 2 kernel threads, one sleeping for the entireinterval and one waiting on terminal input for the interval, the process wait


384

percent values will be 50% on Sleep and 50% on Terminal. The kernelthread wait values will be 100% on Sleep for the first kernel thread and100% on Terminal for the second kernel thread.


PROC_GRAPHICS_WAIT_PCT_CUMTHREAD_GRAPHICS_WAIT_PCT_CUMThe percentage of time the process or kernel thread was blocked ongraphics (waiting for graphics operations to complete) over the cumulativecollection time.








385

PROC_GRAPHICS_WAIT_TIMETHREAD_GRAPHICS_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked ongraphics (waiting for their graphics operations to complete) during theinterval.


PROC_GRAPHICS_WAIT_TIME_CUMTHREAD_GRAPHICS_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked ongraphics (waiting for their graphics operations to complete) over thecumulative collection time.



PROC_GROUP_IDTHREAD_GROUP_IDThe effective group ID number of the process.


PROC_GROUP_NAMETHREAD_GROUP_NAMEThe group name (from /etc/group) of a process.

The group identifier is obtained from searching the /etc/passwd file usingthe user ID (uid) as a key. Therefore, if more than one account is listed in/etc/passwd with the same user ID (uid) field, the first one is used. If no


386

entry can be found for the user ID in /etc/passwd, the group name is theuid number. If no matching entry in /etc/group can be found, the group IDis returned as the group name.


PROC_INODE_WAIT_PCTTHREAD_INODE_WAIT_PCTThe percentage of time the process or kernel thread was blocked onINODE (waiting for an inode to be updated or to become available) duringthe interval.






PROC_INODE_WAIT_PCT_CUMTHREAD_INODE_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon INODE (waiting for an inode to be updated or to become available)over the cumulative collection time.



387






PROC_INODE_WAIT_TIMETHREAD_INODE_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onINODE (waiting for an inode to be updated or to become available) duringthe interval.


PROC_INODE_WAIT_TIME_CUMTHREAD_INODE_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onINODE (waiting for an inode to be updated or to become available) overthe cumulative collection time.



388



PROC_INTERRUPTSTHREAD_INTERRUPTSThe number of interrupts during the interval.

PROC_INTERRUPTS_CUMTHREAD_INTERRUPTS_CUMThe number of interrupts over the cumulative collection time.


PROC_INTERVALTHREAD_INTERVALThe amount of time in the interval. This is the same value for allprocesses and kernel threads, regardless of whether they were alive forthe entire interval.

Note, calculations such as utilizations or rates are computed using thisstandardized process interval (PROC_INTERVAL), rather than the actualalive time during the interval (PROC_INTERVAL_ALIVE). Thus, if aprocess was only alive for 1 second and used the CPU during its entire life(1 second), but the process sample interval was 5 seconds, it would bereported as using 1/5 or 20% CPU utilization, rather than 100% CPUutilization.

PROC_INTERVAL_ALIVETHREAD_INTERVAL_ALIVEThe number of seconds that the process or kernel thread was alive duringthe interval. This may be less than the time of the interval if the process orkernel thread was new or died during the interval.


389

PROC_INTERVAL_CUMTHREAD_INTERVAL_CUMThe amount of time over the cumulative collection time.


PROC_IO_BYTETHREAD_IO_BYTEThe total number of KBs (unless otherwise specified) physically read orwritten by a process or kernel thread directly or indirectly during theinterval.

Indirect IOs include paging and deactivation/reactivation activity done bythe kernel on behalf of the process or kernel thread.

Direct IOs include disk, terminal, tape, and network IO, but exclude allNFS traffic.


PROC_IO_BYTE_CUMTHREAD_IO_BYTE_CUMThe total number of KBs (unless otherwise specified) physically read orwritten by a process or kernel thread directly or indirectly over thecumulative collection time.






390

PROC_IO_BYTE_RATETHREAD_IO_BYTE_RATEThe average number of KBs (unless otherwise specified) physically reador written by a process or kernel thread directly or indirectly during theinterval.




PROC_IO_BYTE_RATE_CUMTHREAD_IO_BYTE_RATE_CUMThe average number of KBs per second physically read or written by aprocess or kernel thread directly or indirectly over the cumulative collectiontime.





PROC_IPC_SUBSYSTEM_WAIT_PCTTHREAD_IPC_SUBSYSTEM_WAIT_PCTThe percentage of time the process or kernel thread was blocked on theInterProcess Communication (IPC) subsystems (waiting for itsinterprocess communication activity to complete) during the interval. This


391

is the sum of processes or kernel threads in the IPC, MSG, SEM, PIPE,SOCKT (that is, sockets) and STRMS (that is, streams IO) wait states.






PROC_IPC_SUBSYSTEM_WAIT_PCT_CUMTHREAD_IPC_SUBSYSTEM_WAIT_PCT_CUMThe percentage of time process or kernel thread was blocked on theInterProcess Communication (IPC) subsystems (waiting for itsinterprocess communication activity to complete) over the cumulativecollection time. This is the sum of processes or kernel threads in the IPC,MSG, SEM, PIPE, SOCKT (that is, sockets) and STRMS (that is, streamsIO) wait states.





392




PROC_IPC_SUBSYSTEM_WAIT_TIMETHREAD_IPC_SUBSYSTEM_WAIT_TIMEThe time, in seconds, the process or kernel thread was blocked on theInterProcess Communication (IPC) subsystems (waiting for itsinterprocess communication activity to complete) during the interval. Thisis the sum of processes or kernel threads in the IPC, MSG, SEM, PIPE,SOCKT (that is, sockets) and STRMS (that is, streams IO) wait states.


PROC_IPC_SUBSYSTEM_WAIT_TIME_CUMTHREAD_IPC_SUBSYSTEM_WAIT_TIME_CUMThe time, in seconds, the process or kernel thread was blocked on theInterProcess Communication (IPC) subsystems (waiting for itsinterprocess communication activity to complete) over the cumulativecollection time. This is the sum of processes or kernel threads in the IPC,MSG, SEM, PIPE, SOCKT (that is, sockets) and STRMS (that is, streamsIO) wait states.


On a threaded operating system, such as HP-UX 11.0 and beyond,process wait time is calculated by summing the wait times of its kernel


393

threads. If this metric is reported for a kernel thread, the value is the waittime of that single kernel thread. If this metric is reported for a process,the value is the sum of the wait times of all of its kernel threads. Alivekernel threads and kernel threads that have died during the interval areincluded in the summation. For multi-threaded processes, the wait timescan exceed the length of the measurement interval.

PROC_IPC_WAIT_PCTTHREAD_IPC_WAIT_PCTThe percentage of time the process or kernel thread was blocked on IPC(waiting for interprocess communication calls to complete) during theinterval.






PROC_IPC_WAIT_PCT_CUMTHREAD_IPC_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon IPC waiting for interprocess communication calls to complete over thecumulative collection time.



394






PROC_IPC_WAIT_TIMETHREAD_IPC_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onInterProcess Communication (IPC) (waiting for its interprocesscommunication calls to complete) during the interval.


PROC_IPC_WAIT_TIME_CUMTHREAD_IPC_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onInterProcess Communication (IPC) (waiting for its interprocesscommunication calls to complete) over the cumulative collection time.



395



PROC_JOBCTL_WAIT_PCTTHREAD_JOBCTL_WAIT_PCTThe percentage of time during the interval the process or kernel threadwas blocked on job control (having been stopped with the job controlfacilities) during the interval. Job control waits include waiting at a debugbreakpoint, as well as being blocked attempting to write (from background)to a terminal which has the “stty tostop” option set.







396

PROC_JOBCTL_WAIT_PCT_CUMTHREAD_JOBCTL_WAIT_PCT_CUMThe percentage of time the process or kernel thread was blocked on jobcontrol (having been stopped with the job control facilities) over thecumulative collection time. Job control waits include waiting at a debugbreakpoint, as well as being blocked attempting to write (from background)to a terminal which has the “stty tostop” option set.







PROC_JOBCTL_WAIT_TIMETHREAD_JOBCTL_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked on jobcontrol (having been stopped with the job control facilities) during theinterval. Job control waits include waiting at a debug breakpoint, as wellas being blocked attempting to write (from background) to a terminal whichhas the “stty tostop” option set.



397


PROC_JOBCTL_WAIT_TIME_CUMTHREAD_JOBCTL_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked on jobcontrol (having been stopped with the job control facilities) over thecumulative collection time. Job control waits include waiting at a debugbreakpoint, as well as being blocked attempting to write (from background)to a terminal which has the “stty tostop” option set.



PROC_LAN_WAIT_PCTTHREAD_LAN_WAIT_PCTThe percentage of time the process or kernel thread was blocked on LAN(waiting for IO over the LAN to complete) during the interval.




For example, if a process has 2 kernel threads, one sleeping for the entireinterval and one waiting on terminal input for the interval, the process wait


398

percent values will be 50% on Sleep and 50% on Terminal. The kernelthread wait values will be 100% on Sleep for the first kernel thread and100% on Terminal for the second kernel thread.


PROC_LAN_WAIT_PCT_CUMTHREAD_LAN_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon LAN (waiting for IO over the LAN to complete) over the cumulativecollection time.








399

PROC_LAN_WAIT_TIMETHREAD_LAN_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onLAN (waiting for IO over the LAN to complete) during the interval.


PROC_LAN_WAIT_TIME_CUMTHREAD_LAN_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onLAN (waiting for IO over the LAN to complete) over the cumulativecollection time.



PROC_MAJOR_FAULTTHREAD_MAJOR_FAULTNumber of major page faults for this process or kernel thread during theinterval.

Major page faults and minor page faults are a subset of vfaults (virtualfaults). Stack and heap accesses can cause vfaults, but do not result in adisk page having to be loaded into memory.

PROC_MAJOR_FAULT_CUMTHREAD_MAJOR_FAULT_CUMNumber of major page faults for this process or kernel thread over thecumulative collection time.


400



PROC_MEM_PRIVATE_RESTHREAD_MEM_PRIVATE_RESThe size (in KB) of resident memory of private regions only, such as dataand stack, currently consumed by this process.

The metric is initialized only when the menu option “Process MemoryRegion” is activated for the process. A value of “na” is displayedotherwise.


PROC_MEM_RESTHREAD_MEM_RESThe size (in KB) of resident memory for the process. This consists of text,data, stack, as well as the process' portion of shared memory regions(such as, shared libraries, text segments, and shared data).

Resident memory (RSS) is calculated as

RSS = sum of private region pages +

(sum of shared region pages /

number of references)


This value is only updated when a process uses CPU. Thus, undermemory pressure, this value may be higher than the actual amount ofresident memory for processes which are idle.

On HP-UX 10.20, the kernel instrumentation doubles the reported size ofprivate regions. To compensate for this, the total reported RSS for eachprocess is halved.

On HP-UX 11.0 and beyond, this metric accurately reports the residentmemory for the process.

Note, a value of “na” may be shown for the swapper process.



401

PROC_MEM_RES_HIGHTHREAD_MEM_RES_HIGHThe largest value of resident memory (in KB) during its lifetime.


PROC_MEM_SHARED_RESTHREAD_MEM_SHARED_RESThe size (in KB) of resident memory of shared regions only, such asshared text, shared memory, and shared libraries. This value is notaffected by the reference count.

A value of “na” is displayed when this information is unobtainable.


PROC_MEM_VFAULT_COUNTTHREAD_MEM_VFAULT_COUNTThe number of times the CPU handled vfaults on behalf of this process orkernel thread during the interval. Major page faults and minor page faultsare a subset of vfaults (virtual faults). Stack and heap accesses cancause vfaults, but do not result in a disk page having to be loaded intomemory.



PROC_MEM_VFAULT_COUNT_CUMTHREAD_MEM_VFAULT_COUNT_CUMThe number of times the CPU handled vfaults on behalf of this process orkernel thread over the cumulative collection time. Major page faults andminor page faults are a subset of vfaults (virtual faults). Stack and heap


402

accesses can cause vfaults, but do not result in a disk page having to beloaded into memory.




PROC_MEM_VIRTTHREAD_MEM_VIRTThis consists of the sum of the virtual set size of all private and sharedmemory regions used by this process. This metric is not affected by thereference count for those regions which are shared.

Note, a value of “na” may be shown for the swapper process.


PROC_MEM_WAIT_PCTTHREAD_MEM_WAIT_PCTThe percentage of time the process or kernel thread was blocked onmemory (waiting for memory resources to become available) during theinterval.





403



PROC_MEM_WAIT_PCT_CUMTHREAD_MEM_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon memory (waiting for memory resources to become available) over thecumulative collection time.








404

PROC_MEM_WAIT_TIMETHREAD_MEM_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onVM (waiting for virtual memory resources to become available) during theinterval.


PROC_MEM_WAIT_TIME_CUMTHREAD_MEM_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onVM (waiting for virtual memory resources to become available) over thecumulative collection time.



PROC_MINOR_FAULTTHREAD_MINOR_FAULTNumber of minor page faults for this process or kernel thread during theinterval.


PROC_MINOR_FAULT_CUMTHREAD_MINOR_FAULT_CUMNumber of minor page faults for this process or kernel thread over thecumulative collection time.


405



PROC_MSG_RECEIVEDTHREAD_MSG_RECEIVEDThe number of socket messages received by a process or kernel threadduring the interval. This does not include SYSV messages (msgrcv).

PROC_MSG_RECEIVED_CUMTHREAD_MSG_RECEIVED_CUMThe total number of socket messages received by a process or kernelthread over the cumulative collection time. This does not include SYSVmessages (msgrcv).


PROC_MSG_SENTTHREAD_MSG_SENTThe number of socket messages sent by a process or kernel thread duringthe interval. This does not include SYSV messages (msgsnd).

PROC_MSG_SENT_CUMTHREAD_MSG_SENT_CUMThe total number of socket messages sent by a process or kernel threadover the cumulative collection time. This does not include SYSVmessages (msgsnd).



406

PROC_MSG_WAIT_PCTTHREAD_MSG_WAIT_PCTThe percentage of time the process or kernel thread was blocked onmessages (waiting for message queue operations to complete) during theinterval.






PROC_MSG_WAIT_PCT_CUMTHREAD_MSG_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon messages (waiting for message queue operations to complete) overthe cumulative collection time.





407




PROC_MSG_WAIT_TIMETHREAD_MSG_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onmessages (waiting for message queue operations to complete) during theinterval.


PROC_MSG_WAIT_TIME_CUMTHREAD_MSG_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onmessages (waiting for message queue operations to complete) over thecumulative collection time.




408


PROC_NFS_WAIT_PCTTHREAD_NFS_WAIT_PCTThe percentage of time the process or kernel thread was blocked on NFS(waiting for network file system IO to complete) during the interval.






PROC_NFS_WAIT_PCT_CUMTHREAD_NFS_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon NFS (waiting for network file system IO to complete) over thecumulative collection time.




409





PROC_NFS_WAIT_TIMETHREAD_NFS_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onNFS (waiting for its network file system IO to complete) during the interval.


PROC_NFS_WAIT_TIME_CUMTHREAD_NFS_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onNFS (waiting for its network file system IO to complete) over thecumulative collection time.


On a threaded operating system, such as HP-UX 11.0 and beyond,process wait time is calculated by summing the wait times of its kernelthreads. If this metric is reported for a kernel thread, the value is the wait


410

time of that single kernel thread. If this metric is reported for a process,the value is the sum of the wait times of all of its kernel threads. Alivekernel threads and kernel threads that have died during the interval areincluded in the summation. For multi-threaded processes, the wait timescan exceed the length of the measurement interval.

PROC_NICE_PRITHREAD_NICE_PRIThe nice priority for the process or kernel thread when it was lastdispatched. The value is a bias used to adjust the priority for the processor kernel thread.

The value ranges from 0 to 39. A higher value causes a process or kernelthread to be dispatched less.


PROC_NONDISK_LOGL_READTHREAD_NONDISK_LOGL_READThe number of non-disk logical reads (that is, calls to read(2)) made by aprocess or kernel thread during the interval.

“Non-disk” devices include terminals, tapes, and so forth on the local orremote machine.


PROC_NONDISK_LOGL_READ_CUMTHREAD_NONDISK_LOGL_READ_CUMThe number of non-disk logical reads (that is, calls to read(2)) made by aprocess or kernel thread over the cumulative collection time.



On a threaded operating system, such as HP-UX 11.0 and beyond,process usage of a resource is calculated by summing the usage of thatresource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metric


411

is reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_NONDISK_LOGL_WRITETHREAD_NONDISK_LOGL_WRITEThe number of non-disk logical writes (that is, calls to write(2)) made by aprocess or kernel thread during the interval.



PROC_NONDISK_LOGL_WRITE_CUMTHREAD_NONDISK_LOGL_WRITE_CUMThe number of non-disk logical writes (that is, calls to write(2)) made by aprocess or kernel thread over the cumulative collection time.




PROC_NONDISK_PHYS_READTHREAD_NONDISK_PHYS_READThe number of physical non-disk reads made by a process or kernelthread during the interval to buffered/block devices, such as a tape drive.

On a threaded operating system, such as HP-UX 11.0 and beyond,process usage of a resource is calculated by summing the usage of thatresource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metric


412

is reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_NONDISK_PHYS_READ_CUMTHREAD_NONDISK_PHYS_READ_CUMThe number of local/remote physical non-disk reads made by a process orkernel thread over the cumulative collection time to buffered/block devices,such as a tape drive.



PROC_NONDISK_PHYS_WRITETHREAD_NONDISK_PHYS_WRITEThe number of local/remote physical non-disk writes made by a process orkernel thread during the interval to buffered/block devices, such as a tapedrive.


PROC_NONDISK_PHYS_WRITE_CUMTHREAD_NONDISK_PHYS_WRITE_CUMNumber of local/remote physical non-disk writes made by a process orkernel thread over the cumulative collection time to buffered/block devices,such as a tape drive.




413

resource by its kernel threads. If this metric is reported for a kernel thread,the value is the resource usage by that single kernel thread. If this metricis reported for a process, the value is the sum of the resource usage by allof its kernel threads. Alive kernel threads and kernel threads that havedied during the interval are included in the summation.

PROC_OPENTHREAD_OPENThe number of file, socket, or pipe opens made by the process or kernelthread during the interval. This corresponds to the number of open(2)system calls.


PROC_OPEN_CUMTHREAD_OPEN_CUMThe number of file, socket, or pipe opens made by the process or kernelthread over the cumulative collection time. This corresponds to thenumber of open(2) system calls.



PROC_OTHER_IO_WAIT_PCTTHREAD_OTHER_IO_WAIT_PCTThe percentage of time the process or kernel thread was blocked on “otherIO” during the interval. “Other IO” includes all IO directed at a device(connected to the local computer) which is not a terminal or LAN.Examples of “other IO” devices are local printers, tapes, instruments, anddisks. Time waiting for character (raw) IO to disks is included in thismeasurement. Time waiting for file system buffered IO to disks willtypically been seen as IO or CACHE wait. Time waiting for IO to NFSdisks is reported as NFS wait.




414




PROC_OTHER_IO_WAIT_PCT_CUMTHREAD_OTHER_IO_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon “other IO” over the cumulative collection time. “Other IO” includes allIO directed at a device (connected to the local computer) which is not aterminal or LAN. Examples of “other IO” devices are local printers, tapes,instruments, and disks. Time waiting for character (raw) IO to disks isincluded in this measurement. Time waiting for file system buffered IO todisks will typically been seen as IO or CACHE wait. Time waiting for IO toNFS disks is reported as NFS wait.







415


PROC_OTHER_IO_WAIT_TIMETHREAD_OTHER_IO_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onother IO during the interval. “Other IO” includes all IO directed at a device(connected to the local computer) which is not a terminal or LAN.Examples of “other IO” devices are local printers, tapes, instruments, anddisks. Time waiting for character (raw) IO to disks is included in thismeasurement. Time waiting for file system buffered IO to disks willtypically been seen as IO or CACHE wait. Time waiting for IO to NFSdisks is reported as NFS wait.


PROC_OTHER_IO_WAIT_TIME_CUMTHREAD_OTHER_IO_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked on“other IO” over the cumulative collection time. “Other IO” includes all IOdirected at a device (connected to the local computer) which is not aterminal or LAN. Examples of “other IO” devices are local printers, tapes,instruments, and disks. Time waiting for character (raw) IO to disks isincluded in this measurement. Time waiting for file system buffered IO todisks will typically been seen as IO or CACHE wait. Time waiting for IO toNFS disks is reported as NFS wait.




416


PROC_OTHER_WAIT_PCTTHREAD_OTHER_WAIT_PCTThe percentage of time the process or kernel thread was blocked on other(unknown) activities during the interval. This includes processes or kernelthreads that were started and subsequently suspended before themidaemon was started and have not been resumed, or the block state isunknown.






PROC_OTHER_WAIT_PCT_CUMTHREAD_OTHER_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon other (unknown) activities over the cumulative collection time. Thisincludes processes or kernel threads that were started and subsequentlysuspended before the midaemon was started and have not been resumed,or the block state is unknown.



417






PROC_OTHER_WAIT_TIMETHREAD_OTHER_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onother (unknown) activities during the interval. This includes processes orkernel threads that were started and subsequently suspended before themidaemon was started and have not been resumed, or the block state isunknown.


PROC_OTHER_WAIT_TIME_CUMTHREAD_OTHER_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onother (unknown) activities over the cumulative collection time. Thisincludes processes or kernel threads that were started and subsequently


418

suspended before the midaemon was started and have not been resumed,or the block state is unknown.



PROC_PARENT_PROC_IDTHREAD_PARENT_PROC_IDThe parent process' PID number.


PROC_PIPE_WAIT_PCTTHREAD_PIPE_WAIT_PCTThe percentage of time the process or kernel thread was blocked on PIPE(waiting for pipe communication to complete) during the interval.





For another example, consider the same process as above, with 2 kernelthreads, one of which was created half-way through the interval, and whichthen slept for the remainder of the interval. The other kernel thread waswaiting for terminal input for half the interval, then used the CPU actively


419

for the remainder of the interval. The process wait percent values will be33% on Sleep and 33% on Terminal (each one third of the total alive time).The kernel thread wait values will be 100% on Sleep for the first kernelthread and 50% on Terminal for the second kernel thread.

PROC_PIPE_WAIT_PCT_CUMTHREAD_PIPE_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon PIPE (waiting for pipe communication to complete) over the cumulativecollection time.







PROC_PIPE_WAIT_TIMETHREAD_PIPE_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onPIPE (waiting for pipe communication to complete) during the interval.



420


PROC_PIPE_WAIT_TIME_CUMTHREAD_PIPE_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onPIPE (waiting for pipe communication to complete) over the cumulativecollection time.



PROC_PRITHREAD_PRIThe dispatch priority of a process or kernel thread at the end of theinterval.

Whenever the priority is changed for the selected process or kernel thread,the new value will not be reflected until the process or kernel thread isreactivated if it is currently idle (for example, SLEEPing).

The lower the value, the more the process or kernel thread is likely to bedispatched. Values between zero and 127 are considered to be “real-time” priorities, which the kernel does not adjust. Values above 127 arenormal priorities and are modified by the kernel for load balancing. Somespecial priorities are used in the HP-UX kernel and subsystems fordifferent activities. These values are described in/usr/include/sys/param.h. Priorities less than PZERO 153 are notsignalable.

Note, many network-related programs such as inetd, biod, and rlogind runat priority 154 which is PPIPE. Just because they run at this priority doesnot mean they are using pipes. By examining the open files, you candetermine if a process or kernel thread is using pipes.

For HP-UX 10.0 and later releases, priorities between -32 and -1 can beseen for processes or kernel threads using the Posix Real-timeSchedulers. When specifying a Posix priority, the value entered must be


421

in the range from 0 through 31, which the system then remaps to anegative number in the range of -1 through -32. Refer to the rtsched manpages for more information.


PROC_PRI_WAIT_PCTTHREAD_PRI_WAIT_PCTThe percentage of time during the interval the process or kernel threadwas blocked on priority (waiting for its priority to become high enough toget the CPU).






PROC_PRI_WAIT_PCT_CUMTHREAD_PRI_WAIT_PCT_CUMThe percentage of time the process or kernel thread was blocked onpriority over the cumulative collection time.


422







PROC_PRI_WAIT_TIMETHREAD_PRI_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onPRI (waiting for its priority to become high enough to get the CPU) duringthe interval.



423

PROC_PRI_WAIT_TIME_CUMTHREAD_PRI_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onPRI (waiting for its priority to become high enough to get the CPU) overthe cumulative collection time.



PROC_PRMIDTHREAD_PRMIDThe PRM Group ID this process is assigned to. The PRM groupconfiguration is kept in the PRM configuration file.


PROC_PROC_IDTHREAD_PROC_IDThe PID number of a process, used by the kernel to identify a process.


PROC_PROC_NAMETHREAD_PROC_NAMEThe process program name. It is limited to 16 characters. This is derivedfrom the 1st parameter to the exec(2) system call.


PROC_REGION_FILENAMEThe file path that corresponds to the front store file of a memory region.For text and data regions, this is the name of the program; for sharedlibraries it is the library name.


424

Certain “special” names are displayed if there is no actual “front store” fora memory region. These special names correspond to the region type (forexample, <stack>). If the name is “<mmap>”, then this is a memory regionwithout “front store,” created by the system call mmap(2).

If the file format includes an inode number, use the program ncheck (1M)to display the filename relative to the mount point. Sometimes files maybe deleted before they are closed. In these cases, the process file tablemay still have the inode even though the file is not actually present and asa result, ncheck will fail.

In the following example, note that the file system name has been includedto avoid the overhead of searching all of the file systems for the inodenumber.

If the following file name was displayed:

<vxfs,/,/dev/root,inode:926>

and then from that display, the following ncheck command was entered:

ncheck -i 926 -F vxfs /dev/root

An output like the following would be generated:

/dev/root:

926 /etc/utmpx

The string for an inode is as follows:

<www,xxx,yyy,zzz,inode:nnnn>

where:

www: Is the file type:

“reg” - Regular file

“dir” - Directory file

“blk” - Block device

“chr” - Character device

“lnk” - Soft file link

“sock” - Socket

“fifo” - FIFO

xxx: Is the file domain; such as,

“ufs” - UNIX file system

“nfs” - NFS

“vxfs” - Veritas file system

yyy: Is the mount point.

zzz: Is the file system.

On HP-UX 11.0, the file type is not shown since it will always be “reg” forregular files, which are mmappable.


425

If a program is “hard linked” (that is, two directories pointing to the sameinode), then a different name may be reported for the text and data regionsthan is actually running. This often happens with the HPTERM program,which is often hard-linked to HELPVIEW. Use the “-i” option of the “ls”command to see the inode numbers.

PROC_REGION_LOCKEDThe amount of memory (in KBs unless otherwise indicated) that is locked.Memory is typically “locked” by calls to plock(2), datalock(3C), orshmctl(2). In addition to the number of pages locked, this metric includesthe number of bytes (rounded up to the nearest page) used by the kernelto store the virtual memory structures allocated to track the pages in thememory region. Hence if the region pages are locked in memory, thevirtual memory management structures for that region must also belocked. As a result, one could see the number of bytes locked exceed thevirtual memory size of a region.

This metric is currently unavailable on HP-UX 11.0.

PROC_REGION_PAGE_COUNT_1_4KBThe number of pages of size 4KB allocated to this memory region.








PROC_REGION_PAGE_COUNT_5_1MBThe number of pages of size 1MB allocated to this memory region.



426







PROC_REGION_PAGE_SIZE_HINTThe recommended or default size for pages allocated to this memoryregion. The chatr(1) command can be used to change the page size hintrequested for a program's text and data regions.


PROC_REGION_PRIVATE_SHARED_FLAGA text indicator of either private memory (Priv) or shared (Shared) for thismemory region. Private memory is only being used by the currentprocess. Shared memory is mapped into the address space of otherprocesses.

PROC_REGION_REF_COUNTThe number of processes sharing this memory region.

For private regions this value is 1. For shared regions, this value is thenumber of processes sharing the region.

This metric is currently unavailable on HP-UX 11.0.

PROC_REGION_RESThe size (in KBs unless otherwise indicated) of the resident memoryoccupied by this memory region.

On HP-UX 10.20, this value is affected by the reference count. Therefore,the sum of all resident region sizes is not equal to physical memory sincesome regions are shared.


427

On HP-UX 11.0 and beyond, this value is not affected by the referencecount.


PROC_REGION_RES_DATAThe size (in KBs unless otherwise indicated) of the total resident memoryoccupied by data regions of this process. This value is not affected by thereference count since all data regions are private.

This metric is specific to the process as a whole and will not change itsvalue. If this metric is used in a glance adviser script, only pick up onevalue. Do not sum the values since the same value is shown for allregions.

PROC_REGION_RES_OTHERThe size (in KBs unless otherwise indicated) of the total resident memoryoccupied by regions of this process that are not text, data, stack, or sharedmemory.





PROC_REGION_RES_SHMEMThe size (in KBs unless otherwise indicated) of the total resident memoryoccupied by shared memory regions of this process.



This metric is specific to the process as a whole and will not change itsvalue. If this metric is used in a glance adviser script, only pick up one


428

value. Do not sum the values since the same value is shown for allregions.


In other words, the sum of this value for all processes exceeds the actualmemory occupied since some memory is shared.

PROC_REGION_RES_STACKThe size (in KBs unless otherwise indicated) of the total resident memoryoccupied by stack regions of this process.

Stack regions are always private and will have a reference count of one.The stack, in this case, refers to the kernel stack, not the user stack. For athreaded OS, each kernel thread will have one kernel stack.


PROC_REGION_RES_TEXTThe size (in KBs unless otherwise indicated) of the total resident memoryoccupied by text regions of this process.





In other words, the sum of this value for all processes exceeds the actualmemory occupied since some memory is shared.

PROC_REGION_TYPEA text name for the type of this memory region. It can be one of thefollowing:


429

DATA Data region

GRAPH Frame buffer lock page

IOMAP IO region (iomap)

LIBTXT Shared Library text

LIBDAT Shared Library data

MEMMAP Memory-mapped file,

which includes shared

libraries (text and

data), or memory

created by calls to

mmap(2).

NULLDR Null pointer dereference

shared page (see below).

SIGSTK Signal stack region

STACK Stack region

TEXT Text (that is, code)

UAREA User Area region

UNKNWN Region of unknown type

A whole page is allocated for NULL pointer dereferencing, which isreported as the NULLDR area. If the program is compiled with the “-z”option (which disallows NULL dereferencing), this area is missing.

Shared libraries are accessed as memory mapped files, so that the codewill show up as “MEMMAP/Shared” and data will show up as“MEMMAP/Priv”.

PROC_REGION_VIRTThe size (in KBs unless otherwise indicated) of the virtual memoryoccupied by this memory region.

This value is not affected by the reference count.


PROC_REGION_VIRT_ADDRSThe virtual address of this memory region displayed in hexadecimalshowing the space and offset of the region.

This is a 64-bit (96-bit on a 64-bit OS) hexadecimal value indicating thespace and space offset of the region.

PROC_REGION_VIRT_DATAThe size (in KBs unless otherwise indicated) of the total virtual memoryoccupied by data regions of this process. This value is not affected by thereference count since all data regions are private.


430


PROC_REGION_VIRT_OTHERThe size (in KBs unless otherwise indicated) of the total virtual memoryoccupied by regions of this process that are not text, data, stack, or sharedmemory.




PROC_REGION_VIRT_SHMEMThe size (in KBs unless otherwise indicated) of the total virtual memoryoccupied by shared memory regions of this process.

Note that this memory is shared by other processes and this figure isreported in their metrics also.




PROC_REGION_VIRT_STACKThe size (in KBs unless otherwise indicated) of the total virtual memoryoccupied by stack regions of this process.

Stack regions are always private and will have a reference count of one.The stack, in this case, refers to the kernel stack, not the user stack. For athreaded OS, each kernel thread will have one kernel stack.



431

PROC_REGION_VIRT_TEXTThe size (in KBs unless otherwise indicated) of the total virtual memoryoccupied by text regions of this process. This value is not affected by thereference count.


PROC_RPC_WAIT_PCTTHREAD_RPC_WAIT_PCTThe percentage of time the process or kernel thread was blocked on RPC(waiting for remote procedure calls to complete) during the interval.






PROC_RPC_WAIT_PCT_CUMTHREAD_RPC_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon RPC (waiting for remote procedure calls to complete) over thecumulative collection time.


432







PROC_RPC_WAIT_TIMETHREAD_RPC_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onRPC (waiting for its remote procedure calls to complete) during theinterval.



433

PROC_RPC_WAIT_TIME_CUMTHREAD_RPC_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onRPC (waiting for its remote procedure calls to complete) over thecumulative collection time.



PROC_RUN_TIMETHREAD_RUN_TIMEThe elapsed time since a process or kernel thread started, in seconds.

This metric is less than the interval time if the process or kernel thread wasnot alive during the entire first or last interval.

On a threaded operating system such as HP-UX 11.0 and beyond, thismetric is available for a process or kernel thread.

PROC_SCHEDULERTHREAD_SCHEDULERThe scheduling policy for this process or kernel thread.

The available scheduling policies are:

HPUX - HP-UX normal timeshare

RTPRIO - HP-UX Real-time

FIFO - Posix First In/First Out

RR - Posix Round-Robin

RR2 - Posix Round-Robin with a

per-priority time slice

interval



434

PROC_SEM_WAIT_PCTTHREAD_SEM_WAIT_PCTThe percentage of time the process or kernel thread was blocked onsemaphores (waiting on a semaphore operation to complete) during theinterval.






PROC_SEM_WAIT_PCT_CUMTHREAD_SEM_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon semaphores (waiting on a semaphore operation to complete) over thecumulative collection time.




435





PROC_SEM_WAIT_TIMETHREAD_SEM_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onsemaphores (waiting on a semaphore operation to complete) during theinterval.


PROC_SEM_WAIT_TIME_CUMTHREAD_SEM_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onsemaphores (waiting on a semaphore operation to complete) over thecumulative collection time.




436


PROC_SIGNALTHREAD_SIGNALNumber of signals seen by the current process or kernel thread during thelifetime of the process or kernel thread.

PROC_SIGNAL_CUMTHREAD_SIGNAL_CUMNumber of signals seen by the current process or kernel thread over thecumulative collection time.


PROC_SLEEP_WAIT_PCTTHREAD_SLEEP_WAIT_PCTThe percentage of time the process or kernel thread was blocked onSLEEP (waiting to awaken from sleep system calls) during the interval. Aprocess or kernel thread enters the SLEEP state by putting itself to sleepusing system calls such as sleep, wait, pause, sigpause, sigsuspend, polland select.






437


PROC_SLEEP_WAIT_PCT_CUMTHREAD_SLEEP_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon SLEEP (waiting to awaken from sleep system calls) over thecumulative collection time. A process or kernel thread enters the SLEEPstate by putting itself to sleep using system calls such as sleep, wait,pause, sigpause, sigsuspend, poll and select.








438

PROC_SLEEP_WAIT_TIMETHREAD_SLEEP_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onSLEEP (waiting to awaken from sleep system calls) during the interval. Aprocess or kernel thread enters the SLEEP state by putting itself to sleepusing system calls such as sleep, wait, pause, sigpause, sigsuspend, polland select.


PROC_SLEEP_WAIT_TIME_CUMTHREAD_SLEEP_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onSLEEP (waiting to awaken from sleep system calls) over the cumulativecollection time. A process or kernel thread enters the SLEEP state byputting itself to sleep using system calls such as sleep, wait, pause,sigpause, sigsuspend, poll and select.



PROC_SOCKET_WAIT_PCTTHREAD_SOCKET_WAIT_PCTThe percentage of time the process or kernel thread was blocked onsockets (waiting for their IO to complete) during the interval.


A percentage of time spent in a wait state is calculated as the time akernel thread (or all kernel threads of a process) spent waiting in this state,


439

divided by the alive time of the kernel thread (or all kernel threads of theprocess) during the interval.




PROC_SOCKET_WAIT_PCT_CUMTHREAD_SOCKET_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon sockets (waiting for their IO to complete) over the cumulative collectiontime.






For another example, consider the same process as above, with 2 kernelthreads, one of which was created half-way through the interval, and which


440

then slept for the remainder of the interval. The other kernel thread waswaiting for terminal input for half the interval, then used the CPU activelyfor the remainder of the interval. The process wait percent values will be33% on Sleep and 33% on Terminal (each one third of the total alive time).The kernel thread wait values will be 100% on Sleep for the first kernelthread and 50% on Terminal for the second kernel thread.

PROC_SOCKET_WAIT_TIMETHREAD_SOCKET_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onsockets (waiting for its IO to complete) during the interval.


PROC_SOCKET_WAIT_TIME_CUMTHREAD_SOCKET_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onsockets (waiting for its IO to complete) over the cumulative collection time.



PROC_STARTTIMETHREAD_STARTTIMEThe creation date and time of the process or kernel thread.


441

PROC_STATETHREAD_STATEA text string summarizing the current state of a process or kernel thread,either:

new This is the first interval

the process/kernel thread has

been displayed.

active Process/kernel thread is

continuing.

died Process/kernel thread expired

during the interval.

PROC_STOP_REASONTHREAD_STOP_REASONA text string describing what caused the process or kernel thread to stopexecuting. For example, if the process is waiting for a CPU while higherpriority processes are executing, then its block reason is PRI. A completelist of block reasons follows:

String Reason for Process Block

------------------------------------

CACHE Waiting at the buffer cache

level trying to lock down a

buffer cache structure, or

waiting for an IO operation

to or from a buffer cache to

complete. File system access

will block on IO more often

than CACHE on HP-UX 11.x.

Processes using the file

system to access block

devices will block on CACHE

on HP-UX 10.20.

CDFS Waiting for CD-ROM file

system node structure

allocation or locks while

accessing a CD-ROM device

through the file system.

died Process terminated during

the interval.

DISK Waiting for an IO operation


442

to complete at the logical

device manager or disk

driver level. Waits from

raw disk IO and diagnostic

requests can be seen here.

Buffered IO requests can

also block on DISK, but will

more often be seen waiting

on “IO”. CDFS access will

block on “CDFS”. Virtual

memory activity will block

on “VM”.

GRAPH Waiting for a graphics card

or framebuf semaphore

operation.

INODE Waiting while accessing

an inode structure. This

includes inode gets and

waiting due to inode locks.

IO Waiting for IO to local

disks, printers, tapes, or

instruments to complete

(above the driver, but below

the buffer cache). Both file

system and raw disk access

can block in this state.

CDFS access will block on

“CDFS”. Virtual memory

activity will block on “VM”.

IPC Waiting for a process or

kernel thread event (that

is, waiting for a child to

receive a signal). This

includes both inter and

intra process or kernel

thread operations, such as

IPC locks, kernel thread

mutexes, and database IPC

operations. System V

message queue operations

will block on “MESG”, while

semaphore operations will

block on “SEM”.

JOBCL Waiting for tracing resume,


443

debug resume, or job control

start. A background process

incurs this block when

attempting to write to a

terminal set with “stty

tostop“. On HP-UX 11i,

scheduler activation threads

(user threads) will show

this block.

LAN Waiting for a network IO

completion. This includes

waiting on the LAN hardware

and low level LAN device

driver. It does not include

waiting on the higher level

network software such as the

streams based transport or

NFS, which has its own stop

state.

MESG Waiting for a System V

message queue operation such

as msgrcv or msgsnd.

new Process was created (via the

fork/vfork system calls)

during the interval.

NFS Waiting for a Networked File

System request to complete.

This includes both NFS V2

and V3 requests. This does

not include stops where

kernel threads or deamons

are waiting for a NFS event

or request (such as biod or

nfsd). These will block on

SLEEP to show they are

waiting for some activity.

NONE Zombie process - waiting to

die.

OTHER The process was started

before the midaemon was

started and has not been

resumed, or the block state

is unknown.


444

PIPE Waiting for operations

involving pipes. This

includes opening, closing,

reading, and writing using

pipes. Named pipes will

block on PIPE.

PRI Waiting because a higher

priority process is running,

or waiting for a spinlock or

alpha semaphore.

RPC Waiting for remote procedure

call operations to complete.

This includes both NFS and

DCE RPC requests.

SEM Waiting for a System V

semaphore operation (such as

semop, semget, or semctl) or

waiting for a memory mapped

file semaphore operation

(such as msem_init or

msem_lock).

SLEEP Waiting because the process

put itself to sleep using

system calls such as sleep,

wait, pause, sigpause, poll,

sigsuspend and select. This

is the standard stop reason

for idle system daemons.

SOCKT Waiting for an operation to

complete while accessing a

device through a socket.

This is used primarily in

networking code and includes

all protocols using sockets

(X25, UDP, TCP, and so on).

STRMS Waiting for an operation to

complete while accessing a

“streams” device. This is

the normal stop reason for

kernel threads and daemons

waiting for a streams event.

This includes the network


445

transport and pseudo

terminal IO requests. For

example, waiting for a read

on a streams device or

waiting for an internal

streams synchronization.

SYSTM Waiting for access to a

system resource or lock.

These resources include data

structures from the LVM,

VFS, UFS, JFS, and Disk

Quota subsystems. “SYSTM”

is the “catch-all” wait

state for blocks on system

resources that are not

common enough or long enough

to warrant their own stop

state.

TERM Waiting for a non-streams

terminal transfer (tty or

pty).

VM Waiting for a virtual memory

operation to complete, or

waiting for free memory, or

blocked while creating/

accessing a virtual memory

structure.

For a process or kernel thread currently running, the last reason it wasstopped before obtaining the CPU is shown.

On HP-UX 11.0 and beyond, mikslp.text (located in /opt/perf/lib) containsthe blocking functions and their corresponding block states for use bymidaemon.


PROC_STOP_REASON_FLAGTHREAD_STOP_REASON_FLAGA numeric value for the stop reason. This is used by scopeux instead ofthe ASCII string returned by PROC_STOP_REASON in order to conservespace in the log file.


446


PROC_STREAM_WAIT_PCTTHREAD_STREAM_WAIT_PCTThe percentage of time the process or kernel thread was blocked onstreams IO (waiting for a streams IO operation to complete) during theinterval.







PROC_STREAM_WAIT_PCT_CUMTHREAD_STREAM_WAIT_PCT_CUMThe average percentage of time the process or thread was blocked onstreams IO (waiting for a streams IO operation to complete) over thecumulative collection time.



447








PROC_STREAM_WAIT_TIMETHREAD_STREAM_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onstreams IO (waiting for a streams IO operation to complete) during theinterval.




448

PROC_STREAM_WAIT_TIME_CUMTHREAD_STREAM_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onstreams IO (waiting for a streams IO operation to complete) over thecumulative collection time.




PROC_SWAPTHREAD_SWAPThe number of times the process or kernel thread was deactivated duringthe interval.





449

PROC_SWAP_CUMTHREAD_SWAP_CUMThe number of times the process or kernel thread was deactivated overthe cumulative collection time.





PROC_SYS_WAIT_PCTTHREAD_SYS_WAIT_PCTThe percentage of time the process or kernel thread was blocked onsystem resources during the interval. These resources include datastructures from the LVM, VFS, UFS, JFS, and Disk Quota subsystems.“SYSTM” is the “catch-all” wait state for blocks on system resources thatare not common enough or long enough to warrant their own stop state.



If this metric is reported for a kernel thread, the percentage value is for thatsingle kernel thread. If this metric is reported for a process, the


450

percentage value is calculated with the sum of the wait and alive times ofall of its kernel threads.



PROC_SYS_WAIT_PCT_CUMTHREAD_SYS_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon SYSTM (that is, system resources) over the cumulative collection time.These resources include data structures from the LVM, VFS, UFS, JFS,and Disk Quota subsystems. “SYSTM” is the “catch-all” wait state forblocks on system resources that are not common enough or long enoughto warrant their own stop state.






For another example, consider the same process as above, with 2 kernelthreads, one of which was created half-way through the interval, and whichthen slept for the remainder of the interval. The other kernel thread waswaiting for terminal input for half the interval, then used the CPU actively


451

for the remainder of the interval. The process wait percent values will be33% on Sleep and 33% on Terminal (each one third of the total alive time).The kernel thread wait values will be 100% on Sleep for the first kernelthread and 50% on Terminal for the second kernel thread.

PROC_SYS_WAIT_TIMETHREAD_SYS_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onSYSTM (that is, system resources) during the interval. These resourcesinclude data structures from the LVM, VFS, UFS, JFS, and Disk Quotasubsystems. “SYSTM” is the “catch-all” wait state for blocks on systemresources that are not common enough or long enough to warrant theirown stop state.


PROC_SYS_WAIT_TIME_CUMTHREAD_SYS_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onSYSTM (that is, system resources) over the cumulative collection time.These resources include data structures from the LVM, VFS, UFS, JFS,and Disk Quota subsystems. “SYSTM” is the “catch-all” wait state forblocks on system resources that are not common enough or long enoughto warrant their own stop state.




452

PROC_TERM_IO_WAIT_PCTTHREAD_TERM_IO_WAIT_PCTThe percentage of time the process or kernel thread was blocked onterminal IO (waiting for its terminal IO to complete) during the interval.






PROC_TERM_IO_WAIT_PCT_CUMTHREAD_TERM_IO_WAIT_PCT_CUMThe average percentage of time the process or kernel thread was blockedon terminal IO (waiting for its terminal IO to complete) over the cumulativecollection time.





453




PROC_TERM_IO_WAIT_TIMETHREAD_TERM_IO_WAIT_TIMEThe time, in seconds, that the process or kernel thread was blocked onterminal IO (waiting for its terminal IO to complete) during the interval.


PROC_TERM_IO_WAIT_TIME_CUMTHREAD_TERM_IO_WAIT_TIME_CUMThe time, in seconds, that the process or kernel thread was blocked onterminal IO (waiting for its terminal IO to complete) over the cumulativecollection time.




454

PROC_THREAD_COUNTTHREAD_THREAD_COUNTThe total number of kernel threads for the current process.

PROC_THREAD_IDTHREAD_THREAD_IDThe thread ID number of this kernel thread, used to uniquely identify it.


PROC_TIMETHREAD_TIMEThe time the data for the process or kernel thread was collected, in localtime.

PROC_TOP_CPU_INDEXTHREAD_TOP_CPU_INDEXThe index of the process which consumed the most CPU during theinterval. From this index, the process PID, process name, and CPUutilization can be obtained.

This metric is used by the Performance Tools to index into the Datacollection interface's internal table. This is not a metric that will beinteresting to Tool users.

PROC_TOP_DISK_INDEXTHREAD_TOP_DISK_INDEXThe index of the process which did the most physical IOs during the lastinterval.

Note that NFS mounted disks are not considered in this calculation.

With this index, the PID, process name, and IOs per second can beobtained.

This metric is used by the Performance Tools to index into the Datacollection interface's internal table. This is not a metric that will beinteresting to Tool's users.


455

PROC_TOTAL_WAIT_TIMETHREAD_TOTAL_WAIT_TIMEThe total time, in seconds, that the process or kernel thread spent blockedduring the interval.


PROC_TOTAL_WAIT_TIME_CUMTHREAD_TOTAL_WAIT_TIME_CUMThe total time that the process or kernel thread spent blocked over thecumulative collection time.



PROC_TTYTHREAD_TTYThe controlling terminal for a process. This field is blank if there is nocontrolling terminal. This is the same as the “TTY” field of the pscommand.


PROC_TTY_DEVTHREAD_TTY_DEVThe device number of the controlling terminal for a process.



456

PROC_UIDTHREAD_UIDThe real UID (user ID number) of a process. This is the UID returned fromthe getuid system call.


PROC_USER_NAMETHREAD_USER_NAMEThe login account of a process (from /etc/passwd).

If more than one account is listed in /etc/passwd with the same user ID(uid) field, the first one is used. If an account cannot be found thatmatches the uid field, then the uid number is returned. This would occur ifthe account was removed after a process was started.


PROC_USER_THREAD_IDTHREAD_USER_THREAD_IDThe user thread ID number of the last user thread to execute within thecontext of this process or kernel thread. User threads IDs are used toidentify user-level threads of execution within the context of a process. Aprocess may have one or more user threads even if there is only onekernel thread.


PROC_USRPRITHREAD_USRPRIThe user priority for the process or kernel thread is set by the kernel duringscheduling. This value becomes the actual process or kernel threadpriority once it returns to user mode from kernel mode. The calculation ofthe user priority is based on the process or kernel thread CPU usage andthe nice value.



457

PROC_VOLUNTARY_CSWITCHTHREAD_VOLUNTARY_CSWITCHThe number of times a process or kernel thread has given up the CPUbefore an external event preempted it during the interval.

Examples of voluntary switches include calls to sleep(2) and select(2).


PROC_VOLUNTARY_CSWITCH_CUMTHREAD_VOLUNTARY_CSWITCH_CUMThe number of times a process or kernel thread has given up the CPUbefore an external event preempted it over the cumulative collection time.Examples of voluntary switches include calls to sleep(2) and select(2).



SYSCALL_ACTIVE_CUMThe number of system calls used on the system. All calls used over thecumulative collection time are included in this count.





458

SYSCALL_CALL_COUNTThe number of system calls made to this function during the interval.

They are assessed when the system call stub returns control back to thecalling program/routine.



SYSCALL_CALL_COUNT_CUMThe number of system calls made to this function over the cumulativecollection time.




SYSCALL_CALL_IDThe ID number of this system call. System calls are sequentiallynumbered starting with one.

SYSCALL_CALL_NAMEThe system call name.

SYSCALL_CALL_RATEThe number of system calls per second made to this function during thelast interval.


459

They are assessed when the system call stub returns control back to thecalling program/routine.



SYSCALL_CALL_RATE_CUMThe average number of system calls per second made to this function overthe cumulative collection time.




SYSCALL_CPU_TOTAL_TIMEThe CPU time, in seconds, during the interval spent executing this systemcalls.

SYSCALL_CPU_TOTAL_TIME_CUMThe CPU time, in seconds, over the cumulative collection time spentexecuting this system calls.


SYSCALL_INTERVALThe amount of time in the interval.


460

SYSCALL_INTERVAL_CUMThe amount of time over the cumulative collection time.


TBL_BUFFER_CACHE_AVAILThe size (in KBs unless otherwise specified) of the file system buffercache on the system. These buffers are used for all file system IOoperations, as well as all other block IO operations in the system (exec,mount, inode reading, and some device drivers).

If dynamic buffer cache is enabled, the system allocates a percentage ofavailable memory not less than dbc_min_pct nor more than dbc_max_pct,depending on the system needs at any given time. On systems with astatic buffer cache, this value will remain equal to bufpages, or not lessthan dbc_min_pct nor more than dbc_max_pct.

TBL_BUFFER_CACHE_HIGHThe highest size (in KBs unless otherwise specified) of the buffer cacheused in any one interval over the cumulative collection time.



TBL_BUFFER_CACHE_MAXThe maximum size (in KBs unless otherwise specified) of the buffer cache.On systems with a dynamic buffer cache, the cache does not exceed thislimit.

This corresponds to the kernel configuration parameter “dbc_max_pct”.On systems with a dynamic buffer cache, the cache does not increaseabove this limit. On systems with a fixed buffer cache, the cache size isequal to the value reported, which is based on the dbc_max_pct orbufpages settings.

TBL_BUFFER_CACHE_MINThe minimum size (in KBs unless otherwise specified) of the buffer cache.On systems with a dynamic buffer cache, the cache does not shrink belowthis limit.


461

This corresponds to the kernel configuration parameter “dbc_min_pct”. Onsystems with a dynamic buffer cache, the cache does not shrink below thislimit. On systems with a fixed buffer cache, the cache size is equal to thevalue reported, which is based on the dbc_min_pct or bufpages settings.

TBL_BUFFER_CACHE_USEDThe size (in KBs unless otherwise specified) of the sum of the currentlyused buffers.

This is normally greater than the amount requested due to internalfragmentation of the buffer cache. Since this is a cache, it is normal for itto be filled. The buffer cache is used to stage all block IOs to disk.

In a dynamic buffer cache configuration, this metric is always equal toTBL_BUFFER_CACHE_AVAIL. With dynamic buffer cache, the systemallocates a percentage of available memory not less than dbc_min_pct normore than dbc_max_pct, depending on the system needs at any giventime.

On systems with a static buffer cache, this value will remain equal tobufpages, or not less than dbc_min_pct nor more than dbc_max_pct. Witha static buffer cache, this metric shows the amount of memory within theconfigured size that is actually used.


TBL_BUFFER_HEADER_AVAILThe configured number of headers pointing to buffers in the file systembuffer cache. This can be set by the “nbuf” kernel configurationparameter. nbuf is used to determine the maximum total number ofbuffers on the system.

These are used to manage the buffer cache, which is used for all block IOoperations. When nbuf is zero, this value depends on the “bufpages” sizeof memory (see System Administration Tasks manual). A value of “na”indicates either a dynamic buffer cache configuration, or the nbuf kernelparameter has been left unconfigured and allowed to “float” with thebufpages parameter.

This is not a maximum available value in a fixed buffer cacheconfiguration. Instead, it is the initial configured value. The actual numberof used buffer headers can grow beyond this initial value.

TBL_BUFFER_HEADER_USEDThe number of buffer headers currently in use.

This dynamic value will rarely change once the system boots. During thesystem bootup, the kernel allocates a large number of buffer headers andthe count is likely to stay at that value after the bootup completes. If thevalue increases beyond the initial boot value, it will not decrease. Buffer


462

headers are allocated in kernel memory, not user memory, and therefore,will not decrease.

This value can exceed the available or configured number of bufferheaders in a fixed buffer cache configuration.


TBL_BUFFER_HEADER_UTILThe percentage of buffer headers currently used.

A value of “na” indicates either a dynamic buffer cache configuration, orthe nbuf kernel parameter has been left unconfigured and allowed to “float”with the bufpages parameter.


TBL_BUFFER_HEADER_UTIL_HIGHThe highest percentage of buffer header used in any one interval over thecumulative collection time.


A value of “na” indicates either a dynamic buffer cache configuration, orthe nbuf kernel parameter has been left unconfigured and allowed to “float”with the bufpages parameter.


TBL_DNLC_CACHE_AVAILThe configured number of entries in the incore directory name cache.



High directory name cache hit rates will be seen on systems wherepathname component requests are frequently repeated. For example,


463

when users or applications work in the same directory where theyrepeatedly list or open the same files, cache hit rates will be high.






TBL_FILE_LOCK_AVAILThe configured number of file or record locks that can be allocated on thesystem. Files and/or records are locked by calls to lockf(2).

TBL_FILE_LOCK_USEDThe number of file or record locks currently in use. One file can havemultiple locks. Files and/or records are locked by calls to lockf(2).


TBL_FILE_LOCK_UTILThe percentage of configured file or record locks currently in use.


TBL_FILE_LOCK_UTIL_HIGHThe highest percentage of configured file or record locks that have been inuse during any one interval over the cumulative collection time.



464


TBL_FILE_TABLE_AVAILThe configured maximum number of the file table entries used by thekernel to manage open file descriptors. This is the sum of the “nfile” and“file_pad” values used in kernel generation.

TBL_FILE_TABLE_USEDThe number of entries in the file table currently used by file descriptors.


TBL_FILE_TABLE_UTILThe percentage of file table entries currently used by file descriptors.


TBL_FILE_TABLE_UTIL_HIGHThe highest percentage of entries in the file table used by file descriptorsin any one interval over the cumulative collection time.



TBL_INODE_CACHE_AVAILThe configured number of entries in the incore inode table on the system.The table contains recently closed inodes as well as open inodes.

Most file system activity is done through inodes which are stored on disk.The kernel keeps a memory cache of active and recently accessed inodesto reduce disk IOs. When a file is accessed through a pathname, thekernel converts the pathname to an inode number and obtains the inodeinformation from the cache. If the inode entry is not in the cache, theinode is read from disk into the inode cache.

The number of used entries in the inode cache is usually at or near thecapacity. This does not necessarily indicate that the size (configurable bythe ninode parameter) is too small because the table contains recentlyused inodes and inodes referenced by entries in the directory name cache.


465

When an inode cache entry is required, the kernel first tries to use a freeentry. If one does not exist, inactive entries referenced by the directoryname cache are used. If after freeing inode entries only referenced by thedirectory name cache does not create enough free space, the inode tablecan be then considered full causing the “inode: table is full” message toappear on the console. If this occurs, increase the ninode parameter.Low directory name cache hit ratios may also indicate an underconfiguredinode cache (ninode).

The default formula for the ninode size is:

ninode = ((nproc+16+maxusers)+32+

(2*npty)+(4*num_clients))

Inodes are used to store information about files within the file system.Every file has at least two inodes associated with it (one for the directoryand one for the file itself). The information stored in an inode includes theowners, timestamps, size, and an array of indices used to translate logicalblock numbers to physical sector numbers. There is a separate inodemaintained for every view of a file, so if two processes have the same fileopen, they both use the same directory inode, but separate inodes for thefile.

TBL_INODE_CACHE_HIGHThe highest number of inodes that have been used in any one intervalover the cumulative collection time.


The number of used entries in the inode cache is usually at or near thecapacity. This does not necessarily indicate that the size (configurable bythe ninode parameter) is too small because the table contains recentlyused inodes and inodes referenced by entries in the directory name cache.When an inode cache entry is required, the kernel first tries to use a freeentry. If one does not exist, inactive entries referenced by the directoryname cache are used. If after freeing inode entries only referenced by thedirectory name cache does not create enough free space, the inode tablecan be then considered full causing the “inode: table is full” message toappear on the console. If this occurs, increase the ninode parameter.Low directory name cache hit ratios may also indicate an underconfiguredinode cache (ninode).





466



TBL_INODE_CACHE_USEDThe number of “non-free” inodes currently used.

Since the inode table contains recently closed inodes as well as openinodes, the table often appears to be fully utilized. When a new entry isneeded, one can usually be found by reusing one of the recently closedinode entries.


The number of used entries in the inode cache is usually at or near thecapacity. This does not necessarily indicate that the size (configurable bythe ninode parameter) is too small because the table contains recentlyused inodes and inodes referenced by entries in the directory name cache.When an inode cache entry is required, the kernel first tries to use a freeentry. If one does not exist, inactive entries referenced by the directoryname cache are used. If after freeing inode entries only referenced by thedirectory name cache does not create enough free space, the inode tablecan be then considered full causing the “inode: table is full” message toappear on the console. If this occurs, increase the ninode parameter.Low directory name cache hit ratios may also indicate an underconfiguredinode cache (ninode).





TBL_MSG_BUFFER_AVAILThe maximum achievable size (in KBs unless otherwise specified) of themessage queue buffer pool on the system.

Each message queue can contain many buffers which are createdwhenever a program issues a msgsnd(2) call. Each of these buffers isallocated from this buffer pool. Also refer to ipcs(1).


467

This value is determined by taking the product of the three kernelconfiguration variables “msgseg”, “msgssz” and “msgmni”.

TBL_MSG_BUFFER_HIGHThe largest size (in KBs unless otherwise specified) of the messagequeues in any one interval over the cumulative collection time.



TBL_MSG_BUFFER_USEDThe current total size (in KBs unless otherwise specified) of all IPCmessage buffers. These buffers are created by msgsnd(2) calls andreleased by msgrcv(2) calls. This field corresponds to the CBYTES field ofthe “ipcs -qo” command.


TBL_MSG_TABLE_AVAILThe configured maximum number of message queues that can beallocated on the system. A message queue is allocated by a programusing the msgget(2) call. Also refer to ipcs(1).

TBL_MSG_TABLE_USEDThe number of message queues currently in use. A message queue isallocated by a program using the msgget(2) call. See ipcs(1) to list themessage queues.


TBL_MSG_TABLE_UTILThe percentage of configured message queues currently in use.



468

TBL_MSG_TABLE_UTIL_HIGHThe highest percentage of configured message queues that have been inuse during any one interval over the cumulative collection time.



TBL_PROC_TABLE_AVAILThe configured maximum number of the proc table entries used by thekernel to manage processes. This number includes both free and usedentries. This is set by the NPROC value during system generation.

TBL_PROC_TABLE_USEDThe number of entries in the proc table currently used by processes.


TBL_PROC_TABLE_UTILThe percentage of proc table entries currently used by processes.


TBL_PROC_TABLE_UTIL_HIGHThe highest percentage of entries in the proc table used by processes inany one interval over the cumulative collection time.



TBL_PTY_AVAILThe configured number of entries used by the pseudo-teletype driver onthe system. This limits the number of pty logins possible.

For HP-UX 10.20, rlogin switched to streams devices, and for HP-UX 11.0and beyond, both telnet and rlogin use streams devices.


469

TBL_PTY_USEDThe number of pseudo-teletype driver (pty) entries currently in use.


TBL_PTY_UTILThe percentage of configured pseudo-teletype driver (pty) entries currentlyin use.


TBL_PTY_UTIL_HIGHThe highest percentage of configured pseudo-teletype driver (pty) entriesin use during any one interval over the cumulative collection time.



TBL_SEM_TABLE_AVAILThe configured number of semaphore identifiers (sets) that can beallocated on the system.

TBL_SEM_TABLE_USEDThe number of semaphore identifiers currently in use. A semaphoreidentifier is allocated by a program using the semget(2) call. See ipcs(1)to list semaphores.


TBL_SEM_TABLE_UTILThe percentage of configured semaphores identifiers currently in use.



470

TBL_SEM_TABLE_UTIL_HIGHThe highest percentage of configured semaphore identifiers that havebeen in use during any one interval over the cumulative collection time.



TBL_SHMEM_AVAILThe maximum achievable size (in MB unless otherwise specified) of theshared memory pool on the system.

This is a theoretical maximum determined by multiplying the configuredmaximum number of shared memory entries (shmmni) by the maximumsize of each shared memory segment (shmmax). Your system may nothave enough virtual memory to actually reach this theoretical limit - onecannot allocate more shared memory than the available reserved spaceconfigured for virtual memory.

It should be noted that this value does not include any architecturallimitations. (For example, on a 32-bit kernel, there is an addressing limit of1.75 GB.)

TBL_SHMEM_REQUESTEDThe size (in KBs unless otherwise specified) of the sum of the currentlyrequested shared memory segments.

This may be more than shared memory used if any segments areswapped out. It also may be less than shared memory used due tointernal fragmentation of the shared memory pool.


TBL_SHMEM_TABLE_AVAILThe configured number of shared memory segments that can be allocatedon the system.

TBL_SHMEM_TABLE_USEDThe number of shared memory segments currently in use. A sharedmemory segment is allocated by a program using the shmget(2) call. Alsorefer to ipcs(1).



471

TBL_SHMEM_TABLE_UTILThe percentage of configured shared memory segments currently in use.


TBL_SHMEM_TABLE_UTIL_HIGHThe highest percentage of configured shared memory segments that havebeen in use during any one interval over the cumulative collection time.



TTBIN_TRANS_COUNTThe number of completed transactions in this range during the lastinterval.

TTBIN_TRANS_COUNT_CUMThe number of completed transactions in this range over the cumulativecollection time.


TTBIN_UPPER_RANGEThe upper range (transaction time) for this TT bin.

There are a maximum of nine user-defined transaction response time bins(TTBIN_UPPER_RANGE). The last bin, which is not specified in ttd.conf,is the overflow bin and will always have a value of -2 (overflow). Note thatthe values specified in ttd.conf cannot exceed 2147483.6, which is thenumber of seconds in 24.85 days. If the user specifies any values greaterthan 2147483.6, the numbers reported for those bins or Service LevelObjectives (SLO) will be -2.


472

TT_ABORTTT_CLIENT_ABORTThe number of aborted transactions during the last interval for thistransaction.

TT_ABORT_CUMTT_CLIENT_ABORT_CUMThe number of aborted transactions over the cumulative collection time forthis transaction.


TT_ABORT_WALL_TIMETT_CLIENT_ABORT_WALL_TIMEThe total time, in seconds, of all aborted transactions during the lastinterval for this transaction.

TT_ABORT_WALL_TIME_CUMTT_CLIENT_ABORT_WALL_TIME_CUMThe total time, in seconds, of all aborted transactions over the cumulativecollection time for this transaction class.


TT_APPNOThe registered ARM Application/User ID for this transaction class.

TT_APP_NAMEThe registered ARM Application name.

TT_CACHE_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on CACHE during the interval.


473

Per-transaction performance resource metrics represent an average forall completed instances of the given transaction during the interval.

If there are no completed transaction instances during an interval, thenthere are no resources accounted, even though there may be in-progresstransactions using resources which have not completed. Resourcemetrics for in-progress transactions will be shown in the interval after theycomplete (that is, after the process has called arm_stop).

If there is only one completed transaction instance during an interval, thenthe resources attributed to the transaction will represent the resourcesused by the process between its call to arm_start and arm_stop, even ifarm_start was called before the current interval. Thus, the resource usagetime or wall time per transaction can exceed the current collection intervaltime.

If there are several completed transaction instances during an interval fora given transaction, then the resources attributed to the transaction willrepresent an average for all completed instances during the interval. Toobtain the total accumulated resource consumption for all completedtransactions during an interval, multiply the resource metric by the numberof completed transaction instances during the interval (TT_COUNT).

TT_CACHE_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on CACHE over the cumulative collection time.


Cumulative per-transaction performance resource metrics represent anaverage for all completed instances of the given transaction over thecumulative collection time.

If there are no completed transaction instances over the cumulativecollection time, then there are no resources accounted, even though theremay be in-progress transactions using resources which have notcompleted. Resource metrics for in-progress transactions will be shown inthe interval after they complete (that is, after the process has calledarm_stop).

If there is only one completed transaction instance over the cumulativecollection time, then the resources attributed to the transaction willrepresent the resources used by the process between its call to arm_startand arm_stop, even if arm_start was called before the current interval.Thus, the resource usage time or wall time per transaction can exceed thecurrent collection interval time.

If there are several completed transaction instances over the cumulativecollection time for a given transaction, then the resources attributed to thetransaction will represent an average for all completed instances over thecumulative collection time. To obtain the total accumulated resourceconsumption for all completed transactions over the cumulative collectiontime, multiply the resource metric by the number of completed transactioninstances over the cumulative collection time (TT_COUNT_CUM).


474

TT_CDFS_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on CDFS (waiting in the CD-ROM driver forCompact Disc file system IO to complete) during the interval.





TT_CDFS_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of transactionwas blocked on CDFS (waiting in the CD-ROM driver for Compact Disc filesystem IO to complete) over the cumulative collection time.




If there is only one completed transaction instance over the cumulativecollection time, then the resources attributed to the transaction willrepresent the resources used by the process between its call to arm_startand arm_stop, even if arm_start was called before the current interval.


475

Thus, the resource usage time or wall time per transaction can exceed thecurrent collection interval time.


TT_CLIENT_ADDRESSTT_INSTANCE_CLIENT_ADDRESSThe correlator address. This is the address where the child transactionoriginated.

TT_CLIENT_ADDRESS_FORMATTT_INSTANCE_CLIENT_ADDRESS_FORMATThe correlator address format. This shows the protocol family for theclient network address. Refer to the ARM API Guide for the list anddescription of supported address formats.

TT_CLIENT_CORRELATOR_COUNTThe number of client or child transaction correlators this transaction hasstarted over the cumulative collection time.


TT_CLIENT_TRAN_IDTT_INSTANCE_CLIENT_TRAN_IDA numerical ID that uniquely identifies the transaction class in thiscorrelator.

TT_COUNTTT_CLIENT_COUNTThe number of completed transactions during the last interval for thistransaction.


476

TT_COUNT_CUMTT_CLIENT_COUNT_CUMThe number of completed transactions over the cumulative collection timefor this transaction.


TT_CPU_CSWITCH_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction spent in context switching during the interval.





TT_CPU_CSWITCH_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction spent in context switching over the cumulative collection time.



If there are no completed transaction instances over the cumulativecollection time, then there are no resources accounted, even though theremay be in-progress transactions using resources which have notcompleted. Resource metrics for in-progress transactions will be shown in


477

the interval after they complete (that is, after the process has calledarm_stop).



TT_CPU_INTERRUPT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction spent processing interrupts during the interval.





TT_CPU_INTERRUPT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction spent processing interrupts over the cumulative collection time.



478





TT_CPU_NICE_TIME_PER_TRANThe average time, in seconds, that each niced instance of the transactionwas using the CPU in user mode during the interval. The NICE metricsinclude positive nice value CPU time only. Negative nice value CPU isbroken out into NNICE (negative nice) metrics. Positive nice values rangefrom 20 to 39. Negative nice values range from 0 to 19.






479

TT_CPU_NICE_TIME_PER_TRAN_CUMThe average time, in seconds, that each niced instance of the transactionwas in user mode over the cumulative collection time. The NICE metricsinclude positive nice value CPU time only. Negative nice value CPU isbroken out into NNICE (negative nice) metrics. Positive nice values rangefrom 20 to 39. Negative nice values range from 0 to 19.






TT_CPU_NNICE_TIME_PER_TRANThe average time, in seconds, that each negatively niced instance of thetransaction was using the CPU in user mode during the interval.



If there is only one completed transaction instance during an interval, thenthe resources attributed to the transaction will represent the resourcesused by the process between its call to arm_start and arm_stop, even ifarm_start was called before the current interval. Thus, the resource usage


480

time or wall time per transaction can exceed the current collection intervaltime.


TT_CPU_NNICE_TIME_PER_TRAN_CUMThe average time, in seconds, that each negatively niced instance of thetransaction was in user mode over the cumulative collection time.






TT_CPU_NORMAL_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was in user mode at normal priority during the interval.

Normal priority user mode CPU excludes CPU used at real-time and nicepriorities.



481




TT_CPU_NORMAL_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was in user mode at normal priority over the cumulativecollection time. Normal priority user mode CPU excludes CPU used atreal-time and nice priorities.







482

TT_CPU_REALTIME_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was in user mode at a realtime priority during the interval.





TT_CPU_REALTIME_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was in user mode at a realtime priority over the cumulativecollection time.






483


TT_CPU_SYSCALL_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was in system mode, excluding interrupt or context processing,during the interval.





TT_CPU_SYSCALL_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was in system mode, excluding interrupt or context processing,over the cumulative collection time.





484




TT_CPU_SYS_MODE_TIME_PER_TRANThe average CPU time in system mode in the context of each completedinstance of the transaction during the interval.





TT_CPU_SYS_MODE_TIME_PER_TRAN_CUMThe average CPU time in system mode in the context of each completedinstance of the transaction over the cumulative collection time.



485





TT_CPU_TOTAL_TIME_PER_TRANThe average total CPU time, in seconds, consumed by each completedinstance of the transaction during the interval.

Total CPU time is the sum of the CPU time components for a process orkernel thread, including system, user, context switch, interrupt processing,realtime, and nice utilization values.






486

TT_CPU_TOTAL_TIME_PER_TRAN_CUMThe average total CPU time consumed by each completed instance of thetransaction over the cumulative collection time. CPU time is in secondsunless otherwise specified.






TT_CPU_USER_MODE_TIME_PER_TRANThe average time, in seconds, each completed instance of the transactionwas using the CPU in user mode during the interval. User CPU is the timespent in user mode at a normal priority, at real-time priority, and at a nicepriority.



If there is only one completed transaction instance during an interval, thenthe resources attributed to the transaction will represent the resourcesused by the process between its call to arm_start and arm_stop, even if


487

arm_start was called before the current interval. Thus, the resource usagetime or wall time per transaction can exceed the current collection intervaltime.


TT_CPU_USER_MODE_TIME_PER_TRAN_CUMThe average time, in seconds, each completed instance of the transactionwas using the CPU in user mode over the cumulative collection time. UserCPU is the time spent in user mode at a normal priority, at real-timepriority, and at a nice priority.






TT_DISK_FS_READ_PER_TRANThe average number of file system physical disk reads made by eachcompleted instance of the transaction during the interval. Only local disksare counted in this measurement. NFS devices are excluded.


488







TT_DISK_FS_READ_PER_TRAN_CUMThe average number of file system physical disk reads made by eachcompleted instance of the transaction over the cumulative collection time.Only local disks are counted in this measurement. NFS devices areexcluded.






489




TT_DISK_FS_WRITE_PER_TRANThe average number of file system physical disk writes made by eachcompleted instance of the transaction during the interval. Only local disksare counted in this measurement. NFS devices are excluded.








490


TT_DISK_FS_WRITE_PER_TRAN_CUMThe average number of file system physical disk writes made by eachcompleted instance of the transaction over the cumulative collection time.Only local disks are counted in this measurement. NFS devices areexcluded.










491

TT_DISK_LOGL_IO_PER_TRANThe average number of logical IOs made by (or for) each completedinstance of the transaction during the interval. NFS mounted disks are notincluded in this list.







TT_DISK_LOGL_IO_PER_TRAN_CUMThe average number of logical IOs made by (or for) each completedinstance of the transaction over the cumulative collection time. NFSmounted disks are not included in this list.





492





TT_DISK_LOGL_READ_PER_TRANThe average number of disk logical reads made by each completedinstance of the transaction during the interval. Calls destined for NFSmounted files are not counted.





If there are several completed transaction instances during an interval fora given transaction, then the resources attributed to the transaction willrepresent an average for all completed instances during the interval. Toobtain the total accumulated resource consumption for all completed


493

transactions during an interval, multiply the resource metric by the numberof completed transaction instances during the interval (TT_COUNT).

TT_DISK_LOGL_READ_PER_TRAN_CUMThe average number of disk logical reads made by each completedinstance of the transaction over the cumulative collection time. Callsdestined for NFS mounted files are not counted.







TT_DISK_LOGL_WRITE_PER_TRANAverage number of disk logical writes made by each completed instanceof the transaction during the interval. Calls destined for NFS mounted filesare not counted.



494





TT_DISK_LOGL_WRITE_PER_TRAN_CUMAverage number of disk logical writes made by each completed instanceof the transaction over the cumulative collection time. Calls destined forNFS mounted files are not counted.







495


TT_DISK_PHYS_IO_PER_TRANThe average number of physical disk IOs per second made by eachcompleted instance of the transaction during the interval.

For transactions which run for less than the measurement interval, thismetric is normalized over the measurement interval. For example, atransaction ran for 1 second and did 50 IOs during its life. If themeasurement interval is 5 seconds, it is reported as having done 10 IOsper second. If the measurement interval is 60 seconds, it is reported ashaving done 50/60 or 0.83 IOs per second.







TT_DISK_PHYS_IO_PER_TRAN_CUMThe average number of physical disk IOs per second made by eachcompleted instance of the transaction over the cumulative collection time.


496


For transactions which run for less than the measurement interval, thismetric is normalized over the measurement interval. For example, atransaction ran for 1 second and did 50 IOs during its life. If themeasurement interval is 5 seconds, it is reported as having done 10 IOsper second. If the measurement interval is 60 seconds, it is reported ashaving done 50/60 or 0.83 IOs per second.







TT_DISK_PHYS_READ_PER_TRANThe average number of physical reads made by (or for) each completedinstance of the transaction during the last interval.




497




TT_DISK_PHYS_READ_PER_TRAN_CUMThe average number of physical reads made by (or for) each completedinstance of the transaction over the cumulative collection time.






If there are several completed transaction instances over the cumulativecollection time for a given transaction, then the resources attributed to thetransaction will represent an average for all completed instances over thecumulative collection time. To obtain the total accumulated resourceconsumption for all completed transactions over the cumulative collection


498

time, multiply the resource metric by the number of completed transactioninstances over the cumulative collection time (TT_COUNT_CUM).

TT_DISK_PHYS_WRITE_PER_TRANThe average number of physical writes made by (or for) each completedinstance of the transaction during the last interval.







TT_DISK_PHYS_WRITE_PER_TRAN_CUMThe average number of physical writes made by (or for) each completedinstance of the transaction over the cumulative collection time.





499





TT_DISK_RAW_READ_PER_TRANThe average number of raw reads made for each completed instance ofthe transaction during the interval.






If there are several completed transaction instances during an interval fora given transaction, then the resources attributed to the transaction willrepresent an average for all completed instances during the interval. To


500

obtain the total accumulated resource consumption for all completedtransactions during an interval, multiply the resource metric by the numberof completed transaction instances during the interval (TT_COUNT).

TT_DISK_RAW_READ_PER_TRAN_CUMThe average number of raw reads made for each completed instance ofthe transaction over the cumulative collection time.








TT_DISK_RAW_WRITE_PER_TRANThe average number of raw writes made for each completed instance ofthe transaction during the interval.



501






TT_DISK_RAW_WRITE_PER_TRAN_CUMThe average number of raw writes made for each completed instance ofthe transaction over the cumulative collection time.






If there is only one completed transaction instance over the cumulativecollection time, then the resources attributed to the transaction willrepresent the resources used by the process between its call to arm_start


502

and arm_stop, even if arm_start was called before the current interval.Thus, the resource usage time or wall time per transaction can exceed thecurrent collection interval time.


TT_DISK_SYSTEM_READ_PER_TRANThe average number of file system management physical disk reads madefor each completed instance of the transaction during the interval.







TT_DISK_SYSTEM_READ_PER_TRAN_CUMThe average number of file system management physical disk reads madefor each instance completed of the transaction over the cumulativecollection time.



503








TT_DISK_SYSTEM_WRITE_PER_TRANThe average number of file system management physical disk writesmade for each completed instance of the transaction during the interval.





If there are no completed transaction instances during an interval, thenthere are no resources accounted, even though there may be in-progress


504

transactions using resources which have not completed. Resourcemetrics for in-progress transactions will be shown in the interval after theycomplete (that is, after the process has called arm_stop).



TT_DISK_SYSTEM_WRITE_PER_TRAN_CUMThe average number of file system management physical disk writesmade for each completed instance of the transaction over the cumulativecollection time.









505


TT_DISK_VM_READ_PER_TRANThe average number of virtual memory reads made for each completedinstance of the transaction during the interval.






TT_DISK_VM_READ_PER_TRAN_CUMThe average number of virtual memory reads made for each completedinstance of the transaction over the cumulative collection time.





506




TT_DISK_VM_WRITE_PER_TRANThe average number of virtual memory writes made for each completedinstance of the transaction during the interval.








507

TT_DISK_VM_WRITE_PER_TRAN_CUMThe average number of virtual memory writes made for each completedinstance of the transaction over the cumulative collection time.








TT_DISK_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on DISK during the interval.




508



TT_DISK_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on DISK over the cumulative collection time.






TT_FAILEDTT_CLIENT_FAILEDThe number of Failed transactions during the last interval for thistransaction name.


509

TT_FAILED_CUMTT_CLIENT_FAILED_CUMThe number of failed transactions over the cumulative collection time forthis transaction name.


TT_FAILED_WALL_TIMETT_CLIENT_FAILED_WALL_TIMEThe total time, in seconds, of all failed transactions during the last intervalfor this transaction name.

TT_FAILED_WALL_TIME_CUMTT_CLIENT_FAILED_WALL_TIME_CUMThe total time, in seconds, of all failed transactions over the cumulativecollection time for this transaction name.


TT_GOLDENRESOURCE_INTERVALThe amount of time in the collection interval.

TT_GOLDENRESOURCE_INTERVAL_CUMThe amount of time over the cumulative collection time.


TT_GRAPHICS_WAIT_TIME_PER_TRANThe average time that each completed instance of the transaction wasblocked on graphics (waiting for their graphics operations to complete)during the interval.


510





TT_GRAPHICS_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on graphics (waiting for their graphics operationsto complete) over the cumulative collection time.





If there are several completed transaction instances over the cumulativecollection time for a given transaction, then the resources attributed to thetransaction will represent an average for all completed instances over thecumulative collection time. To obtain the total accumulated resourceconsumption for all completed transactions over the cumulative collection


511

time, multiply the resource metric by the number of completed transactioninstances over the cumulative collection time (TT_COUNT_CUM).

TT_INFOThe registered ARM Transaction Information for this transaction.

TT_INODE_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on INODE (waiting for an inode to be updated orto become available) during the interval.





TT_INODE_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on INODE (waiting for an inode to be updated orto become available) over the cumulative collection time.





512




TT_INPROGRESS_COUNTThe number of transactions in progress (started, but not stopped) at theend of the interval for this transaction class.

TT_INSTANCE_IDA numerical ID that uniquely identifies this transaction instance at the endof the interval.

TT_INSTANCE_PROC_IDThe ID of the process that started or last updated the transaction instance.

TT_INSTANCE_START_TIMEThe time this transaction instance started.

TT_INSTANCE_STOP_TIMEThe time this transaction instance stopped. If the transaction instance iscurrently active, the value returned will be -1. It will be shown as “na” inGlance and GPM to indicate that the transaction instance did not stopduring the interval.

TT_INSTANCE_THREAD_IDThe ID of the kernel thread that started or last updated the transactioninstance.


513

TT_INSTANCE_UPDATE_COUNTThe number of times this transaction instance called update since the startof this transaction instance.

TT_INSTANCE_UPDATE_TIMEThe time this transaction instance last called update. If the transactioninstance is currently active, the value returned will be -1. It will be shownas “na” in Glance and GPM to indicate that a call to update did not occurduring the interval.

TT_INSTANCE_WALL_TIMEThe elapsed time since this transaction instance was started.

TT_INTERVALTT_CLIENT_INTERVALThe amount of time in the collection interval.

TT_INTERVAL_CUMTT_CLIENT_INTERVAL_CUMThe amount of time over the cumulative collection time.


TT_IPC_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on IPC during the interval.





514


TT_IPC_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on IPC over the cumulative collection time.






TT_JOBCTL_WAIT_TIME_PER_TRANThe average time that each completed instance of the transaction wasblocked on job control (having been stopped with the job control facilities)during the interval. Job control waits include waiting at a debugbreakpoint, as well as being blocked attempting to write (from background)to a terminal which has the “stty tostop” option set.


If there are no completed transaction instances during an interval, thenthere are no resources accounted, even though there may be in-progress


515

transactions using resources which have not completed. Resourcemetrics for in-progress transactions will be shown in the interval after theycomplete (that is, after the process has called arm_stop).



TT_JOBCTL_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on job control (having been stopped with the jobcontrol facilities) over the cumulative collection time. Job control waitsinclude waiting at a debug breakpoint, as well as being blocked attemptingto write (from background) to a terminal which has the “stty tostop” optionset.







516

TT_LAN_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on LAN (waiting for IO over the LAN to complete)during the interval.





TT_LAN_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on LAN (waiting for IO over the LAN to complete)over the cumulative collection time.






517



TT_MEASUREMENT_COUNTThe number of user defined measurements for this transaction class.

TT_MEM_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on memory during the interval.





TT_MEM_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on memory over the cumulative collection time.



518





TT_MSG_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on messages (waiting for message queueoperations to complete) during the interval.






519

TT_MSG_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on messages (waiting for message queueoperations to complete) over the cumulative collection time.






TT_NAMEThe registered transaction name for this transaction.

TT_NFS_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on NFS (waiting for its network file system IO tocomplete) during the interval.




520



TT_NFS_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on NFS (waiting for its network file system IO tocomplete) over the cumulative collection time.






TT_OTHER_IO_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on “other IO” during the interval. “Other IO”includes all IO directed at a device (connected to the local computer)


521

which is not a terminal or LAN. Examples of “other IO” devices are localprinters, tapes, instruments, and disks. Time waiting for character (raw) IOto disks is included in this measurement. Time waiting for file systembuffered IO to disks will typically been seen as IO or CACHE wait. Timewaiting for IO to NFS disks is reported as NFS wait.





TT_OTHER_IO_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on “other IO” over the cumulative collection time.“Other IO” includes all IO directed at a device (connected to the localcomputer) which is not a terminal or LAN. Examples of “other IO” devicesare local printers, tapes, instruments, and disks. Time waiting forcharacter (raw) IO to disks is included in this measurement. Time waitingfor file system buffered IO to disks will typically been seen as IO orCACHE wait. Time waiting for IO to NFS disks is reported as NFS wait.




If there is only one completed transaction instance over the cumulativecollection time, then the resources attributed to the transaction will


522

represent the resources used by the process between its call to arm_startand arm_stop, even if arm_start was called before the current interval.Thus, the resource usage time or wall time per transaction can exceed thecurrent collection interval time.


TT_OTHER_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on other (unknown) activities during the interval.This includes transactions that were started and subsequently suspendedbefore the midaemon was started and have not been resumed, or theblock state is unknown.





TT_OTHER_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on other (unknown) activities over the cumulativecollection time. This includes transactions that were started andsubsequently suspended before the midaemon was started and have notbeen resumed, or the block state is unknown.



523





TT_PIPE_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on PIPE (waiting for pipe communication tocomplete) during the interval.






524

TT_PIPE_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on PIPE (waiting for pipe communication tocomplete) over the cumulative collection time.






TT_PRI_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on priority during the interval.





525


TT_PRI_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on priority over the cumulative collection time.






TT_RESOURCE_INTERVALThe amount of time in the collection interval.

TT_RESOURCE_INTERVAL_CUMThe amount of time over the cumulative collection time.



526

TT_RPC_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on RPC (waiting for its remote procedure calls tocomplete) during the interval.





TT_RPC_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on RPC (waiting for its remote procedure calls tocomplete) over the cumulative collection time.






527



TT_SEM_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on semaphores (waiting on a semaphoreoperation to complete) during the interval.





TT_SEM_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on semaphores (waiting on a semaphoreoperation to complete) over the cumulative collection time.



If there are no completed transaction instances over the cumulativecollection time, then there are no resources accounted, even though there


528

may be in-progress transactions using resources which have notcompleted. Resource metrics for in-progress transactions will be shown inthe interval after they complete (that is, after the process has calledarm_stop).



TT_SLEEP_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on SLEEP (waiting to awaken from sleep systemcalls) during the interval. A transaction enters the SLEEP state by puttingitself to sleep using system calls such as sleep, wait, pause, sigpause,sigsuspend, poll and select.





TT_SLEEP_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on SLEEP (waiting to awaken from sleep systemcalls) over the cumulative collection time. A transaction enters the SLEEP


529

state by putting itself to sleep using system calls such as sleep, wait,pause, sigpause, sigsuspend, poll and select.






TT_SLO_COUNTTT_CLIENT_SLO_COUNTThe number of completed transactions that violated the defined ServiceLevel Objective (SLO) by exceeding the SLO threshold time during theinterval.

TT_SLO_COUNT_CUMTT_CLIENT_SLO_COUNT_CUMThe number of completed transactions that violated the defined ServiceLevel Objective by exceeding the SLO threshold time over the cumulativecollection time.



530

TT_SLO_THRESHOLDThe upper range (transaction time) of the Service Level Objective (SLO)threshold value. This value is used to count the number of transactionsthat exceed this user-supplied transaction time value.

TT_SOCKET_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on sockets (waiting for its IO to complete) duringthe interval.





TT_SOCKET_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on sockets (waiting for its IO to complete) overthe cumulative collection time.





531



TT_STREAM_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on streams IO (waiting for a streams IO operationto complete) during the interval.






TT_STREAM_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on streams IO (waiting for a streams IO operationto complete) over the cumulative collection time.



532






TT_SYS_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on system blocked on SYSTM (that is, systemresources) during the interval. These resources include data structuresfrom the LVM, VFS, UFS, JFS, and Disk Quota subsystems. “SYSTM” isthe “catch-all” wait state for blocks on system resources that are notcommon enough or long enough to warrant their own stop state.




If there are several completed transaction instances during an interval fora given transaction, then the resources attributed to the transaction willrepresent an average for all completed instances during the interval. Toobtain the total accumulated resource consumption for all completed


533

transactions during an interval, multiply the resource metric by the numberof completed transaction instances during the interval (TT_COUNT).

TT_SYS_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on SYSTM (that is, system resources) over thecumulative collection time. These resources include data structures fromthe LVM, VFS, UFS, JFS, and Disk Quota subsystems. “SYSTM” is the“catch-all” wait state for blocks on system resources that are not commonenough or long enough to warrant their own stop state.






TT_TERM_IO_WAIT_TIME_PER_TRANThe average time, in seconds, that each completed instance of thetransaction was blocked on terminal IO (waiting for its terminal IO tocomplete) during the interval.


If there are no completed transaction instances during an interval, thenthere are no resources accounted, even though there may be in-progresstransactions using resources which have not completed. Resource


534

metrics for in-progress transactions will be shown in the interval after theycomplete (that is, after the process has called arm_stop).



TT_TERM_IO_WAIT_TIME_PER_TRAN_CUMThe average time, in seconds, that each completed instance of thetransaction was blocked on terminal IO (waiting for its terminal IO tocomplete) over the cumulative collection time.







535

TT_TOTAL_WAIT_TIME_PER_TRANThe average total time that each completed instance of the transactionspent blocked during the interval.





TT_TOTAL_WAIT_TIME_PER_TRAN_CUMThe average total time that each completed instance of the transactionspent blocked during over the cumulative collection time.





If there are several completed transaction instances over the cumulativecollection time for a given transaction, then the resources attributed to thetransaction will represent an average for all completed instances over the


536

cumulative collection time. To obtain the total accumulated resourceconsumption for all completed transactions over the cumulative collectiontime, multiply the resource metric by the number of completed transactioninstances over the cumulative collection time (TT_COUNT_CUM).

TT_TRAN_IDThe registered ARM Transaction ID for this transaction class as returnedby arm_getid(). A unique transaction id is returned for a uniqueapplication id (returned by arm_init), tran name, and meta data buffercontents.

TT_UNAMEThe registered ARM Transaction User Name for this transaction.

If the arm_init function has NULL for the appl_user_id field, then the username is blank. Otherwise, if “*” was specified, then the user name isdisplayed.

For example, to show the user name for the armsample1 program, use:

appl_id = arm_init(“armsample1”,“*”,0,0,0);

To ignore the user name for the armsample1 program, use:

appl_id = arm_init(“armsample1”,NULL,0,0,0);

TT_UPDATETT_CLIENT_UPDATEThe number of updates during the last interval for this transaction class.This count includes update calls for completed and in progresstransactions.

TT_UPDATE_CUMTT_CLIENT_UPDATE_CUMThe number of updates over the cumulative collection time for thistransaction class. This count includes update calls for completed and inprogress transactions.



537

TT_USER_MEASUREMENT_AVGTT_INSTANCE_USER_MEASUREMENT_AVGTT_CLIENT_USER_MEASUREMENT_AVGIf the measurement type is a numeric or a string, this metric returns “na”.

If the measurement type is a counter, this metric returns the averagecounter differences of the transaction or transaction instance during thelast interval. The counter value is the difference observed from a counterbetween the start and the stop (or last update) of a transaction.

If the measurement type is a gauge, this returns the average of the valuespassed on any ARM call for the transaction or transaction instance duringthe last interval.

TT_USER_MEASUREMENT_COUNTTT_INSTANCE_USER_MEASUREMENT_COUNTTT_CLIENT_USER_MEASUREMENT_COUNTThis returns the total number of times the associated user defined metric(UDM) was sampled during the last interval.

TT_USER_MEASUREMENT_MAXTT_INSTANCE_USER_MEASUREMENT_MAXTT_CLIENT_USER_MEASUREMENT_MAXIf the measurement type is a numeric or a string, this metric returns “na”.

If the measurement type is a counter, this metric returns the highestmeasured counter value over the life of the transaction or transactioninstance. The counter value is the difference observed from a counterbetween the start and the stop (or last update) of a transaction.

If the measurement type is a gauge, this metric returns the highest valuepassed on any ARM call over the life of the transaction or transactioninstance.

TT_USER_MEASUREMENT_MINTT_INSTANCE_USER_MEASUREMENT_MINTT_CLIENT_USER_MEASUREMENT_MINIf the measurement type is a numeric or a string, this metric returns “na”.

If the measurement type is a counter, this metric returns the lowestmeasured counter value over the life of the transaction or transactioninstance. The counter value is the difference observed from a counterbetween the start and the stop (or last update) of a transaction.

If the measurement type is a gauge, this metric returns the lowest valuepassed on any ARM call over the life of the transaction or transactioninstance.


538

TT_USER_MEASUREMENT_NAMETT_INSTANCE_USER_MEASUREMENT_NAMETT_CLIENT_USER_MEASUREMENT_NAMEThe name of the user defined transactional measurement. The length ofthe string complies with the ARM 2.0 standard, which is 44 characters long(there are 43 usable characters since this is a NULL terminated characterstring).

TT_USER_MEASUREMENT_STRING1024_VALUETT_INSTANCE_USER_MEASUREMENT_STRING1024_VALUETT_CLIENT_USER_MEASUREMENT_STRING1024_VALUEThe last value of the user defined measurement of type string 1024.

TT_USER_MEASUREMENT_STRING32_VALUETT_INSTANCE_USER_MEASUREMENT_STRING32_VALUETT_CLIENT_USER_MEASUREMENT_STRING32_VALUEThe last value of the user defined measurement of type string 32.

TT_USER_MEASUREMENT_TYPETT_INSTANCE_USER_MEASUREMENT_TYPETT_CLIENT_USER_MEASUREMENT_TYPEThe type of the user defined transactional measurement.

1 = ARM_COUNTER32

2 = ARM_COUNTER64

3 = ARM_CNTRDIVR32

4 = ARM_GAUGE32

5 = ARM_GAUGE64

6 = ARM_GAUGEDIVR32

7 = ARM_NUMERICID32

8 = ARM_NUMERICID64

9 = ARM_STRING8 (max 8 chars)

10 = ARM_STRING32 (max 32 chars)

11 = ARM_STRING1024 (max 1024 char)


539

TT_USER_MEASUREMENT_VALUETT_INSTANCE_USER_MEASUREMENT_VALUETT_CLIENT_USER_MEASUREMENT_VALUEThe last value of the user defined measurement of type counter, gauge,numeric ID, or string 8. Both 32 and 64 bit numeric types are returned as64 bit values.

TT_WALL_TIMETT_CLIENT_WALL_TIMEThe total time, in seconds, of all transactions completed during the lastinterval for this transaction.

TT_WALL_TIME_CUMTT_CLIENT_WALL_TIME_CUMThe total time, in seconds, of all transactions completed over thecumulative collection time for this transaction.


TT_WALL_TIME_PER_TRANTT_CLIENT_WALL_TIME_PER_TRANThe average transaction time, in seconds, during the last interval for thistransaction.

TT_WALL_TIME_PER_TRAN_CUMTT_CLIENT_WALL_TIME_PER_TRAN_CUMThe average transaction time, in seconds, over the cumulative collectiontime for this transaction.



540

Glossary

alarmA signal that an event has occurred. The signal can be either anotification or an automatically triggered action. The event can be apre-defined threshold that is exceeded, a network node in trouble, andso on. Alarm information can be sent to an OV Performance Manageranalysis system and to HP OpenView and IT/Operations. Alarms canalso be identified in historical log file data.

alarm generatorThe service that handles the communication of alarm information. Itconsists of the agdbserver and the agdb database that it manages. Theagdb database contains a list of OV Performance Manager (PerfView)analysis notes (if any) to which alarms are communicated, and variouson/off flags that are set to define when and where the alarminformation is sent.

alarmdef fileThe file containing the alarm definitions in which alarm conditions arespecified.

alertA message sent when alarm conditions or conditions in an IFstatement have been met.

analysis softwareAnalysis software analyzes system performance data.

The optional OV Performance Manager product provides a centralwindow from which you can monitor, manage, and troubleshoot theperformance of all networked systems in your computing environment,as well as analyze historical data from OV Performance Agent systems.With OV Performance Manager, you view incoming alarms from allmonitored systems on your network and view graphs of a systems'performance data to help you diagnose and resolve performanceproblems quickly.

applicationA user-defined group of related processes or program files.Applications are defined so that performance software can collectperformance metrics for and report on the combined activities of theprocesses and programs.

GlancePlus Dictionary of Performance MetricsGlossary

541

available memoryAvailable memory is that part of physical memory not allocated by thekernel. This includes the buffer cache, user allocated memory, and freememory.

backtrackBacktracking allows the large data structures used by the VirtualMemory Manager (VMM) to be pageable. It is a method of safelyallowing the VMM to handle page faults within its own critical sectionsof code.

Examples of backtracking are:

• A process page faults.

• The VMM attempts to locate the missing page via its External Page table (XPT).

• The VMM page faults due to the required XPT itself having been paged out.

• The VMM safely saves enough information on the stack to restart the process

• at its first fault.

• Normal VMM pagein/out routines are used to recover the missing XPT.

• The required XPT is now present, so the missing page is located and paged-in.

• The process continues normal execution at the original page fault.

bad callA failed NFS server call. Calls fail due to lack of system resources(lack of virtual memory) and network errors.

biodA daemon process responsible for asynchronous block IO on the NFSclient. It is used to buffer read-ahead and write-behind IOs.

block IOBuffered reads and writes. Data is held in the buffer cache, thentransferred in fixed-size blocks. Any hardware device that transmitsand receives data in blocks is a block-mode device. Compare withcharacter mode.

block IO bufferA buffer used to store data being transferred to or from a block-modedevice through file system input and output, as opposed to character-mode or raw-mode devices.

block IO operationAny operation being carried out on a block-mode device (such as read,write, or mount).


542

block sizeThe size of the primary unit of information used for a file system. It isset when a file system is created.

blocked stateThe reason for the last recorded process block. Also called blocked-onstate.

bottleneckA situation that occurs when a system resource is constrained bydemand that exceeds its capability. The resource is said to be"bottlenecked." A bottleneck causes system performance to degrade. Aprimary characteristic of a bottleneck is that it does not occur in allresources at the same time; other resources may instead beunderutilized.

bufferA memory storage area used to temporarily hold code or data untilused for input/output operations.

buffer cacheAn area of memory that mediates between application programs anddisk drives. When a program writes data, it is first placed in the buffercache, then delivered to the disk at a later time. This allows the diskdriver to perform IO operations in batches, minimizing seek time.

buffer headerEntries used by all block IO operations to point to buffers in the filesystem buffer cache.

buffer poolSee buffer cache.

cacheSee buffer cache.

cache efficiencyThe extent to which buffered read and read-ahead requests can besatisfied by data already in the cache.

cache hitRead requests that are satisfied by data already in the buffer cache.See also cache efficiency.


543

character modeThe mode in which data transfers are accomplished byte-by-byte,rather than in blocks. Printers, plotters, and terminals are examples ofcharacter-mode devices. Also known as raw mode. Compare with blockIO.

child processA new process created at another active process' request through a forkor vfork system call. The process making the request becomes theparent process.

clientA system that requests a service from a server. In the context ofdiskless clusters, a client uses the server's disks and has none of itsown. In the context of NFS, a client mounts file systems thatphysically reside on another system (the Network File System server).

clock hand algorithmThe algorithm used by the page daemon to scan pages.

clock hand cycleThe clock hand algorithm used to control paging and to select pages forremoval from system memory. When page faults and/or systemdemands cause the free list size to fall below a certain level, the pagereplacement algorithm starts the clock hand and it cycles through thepage table.

clusterOne or more work stations linked by a local area network (LAN) buthaving only one root file system.

cluster server process(CSPs). A special kernel process that runs in a cluster and handlesrequests from remote cnodes.

cnodeThe client on a diskless system. The term cnode is derived from "clientnode."

collisionOccurs when the system attempts to send a packet at the same timethat another system is attempting a send on the same LAN. The resultis garbled transmissions and both sides have to resubmit the packet.Some collisions occur during normal operation.


544

context switchThe action of the dispatcher (scheduler) changing from running oneprocess to another. The scheduler maintains algorithms for managingprocess switching, mostly directed by process priorities.

CPUCentral Processing Unit. The part of a computer that executesprogram instructions.

CPU entitlementThe percentage of CPU guaranteed to a particular Process ResourceManager (PRM) group when the total system CPU use is at 100%. Thesystem administrator assigns the CPU entitlement for each PRM groupin the PRM configuration file. The minimum entitlement for all groupsis 1%. PRM distributes unused time to groups in proportion to theirCPU entitlements.

CPU queueThe average number of processes in the "run" state awaiting CPUscheduling, which includes processes short waited for IOs. This iscalculated from GBL-RUN-QUEUE and the number of times thismetric is updated. This is also a measure of how busy the system'sCPU resource is.

cyclical redundancy check(CRC). A networking checksum protocol used to detect transmissionerrors.

cylinderThe tracks of a disk accessible from one position of the head assembly.

cylinder groupIn the filesystem, a collection of cylinders on a disk drive groupedtogether for the purpose of localizing information.

The filesystem allocates inodes and data blocks on a per-cylinder-groupbasis.

daemonA process that runs continuously in the background but providesimportant system services.

data classA particular category of data collected by a data collection process.Single-instance data classes, such as the global class, contain a singleset of metrics that appear only once in any data source. Multiple-


545

instance classes, such as the application class, may have manyoccurrences in a single data source, with the same set of metricscollected for each occurrence of the class.

data localityThe location of data relative to associated data. Associated data hasgood data locality if it is located near one another, because accesses arelimited to a small number of pages and the data is more likely to be inmemory. Poor data locality means associated data must be obtainedfrom different data pages.

data pointA specific point in time displayed on a performance graph where datahas been summarized every five, fifteen, or thirty minutes, or everyhour, two hours or one day.

data segmentA section of memory reserved for storing a process' static and dynamicdata.

data sourceA data source consists of one or more classes of data in a single scopeuxor DSI log file set. For example, the default OV Performance Agentdata source, SCOPE, is a scopeux log file set consisting of global data.See also log file set and respository server.

data source integration (DSI)Enables OV Performance Server Agent to receive, log, and detectalarms on dat from external sources such as applications, databases,networks, and other operating systems.

deactivated/reactivated pages outPages from deactivated process regions that are moved from memory tothe swap area. These pages are swapped out only when they areneeded by another active process.

When a process becomes reactivated, the pages are moved from theswap area back to memory.

defaultAn option that is automatically selected or chosen by the system.

deferred packetA deferred packet occurs when the network hardware detects that theLAN is already in use. Rather than incur a collision, the outboundpacket transmission is delayed until the LAN is available.


546

device driverA collection of kernel routines and data structures that handle thelowest levels of input and output between a peripheral device andexecuting processes. Device drivers are part of the UNIX kernel.

device fileA special file that permits direct access to a hardware device.

device swap spaceSpace devoted to swapping.

directory name lookup cacheThe directory name lookup cache (DNLC) is used to cache directory andfile names. When a file is referenced by name, the name must bebroken into its components and each component's inode must be lookedup. By caching the component names, disk IOs are reduced.

disk bandwidth entitlementThe percentage of disk (volume group) bandwidth guaranteed to aparticular PRM group when the total system disk bandwidth use is atits maximum. The system administrator assigns the disk bandwidthentitlement for each PRM group in the PRM configuration file. Theminimum entitlement for groups other than the system group is 1%.PRM distributes unused time to other groups in proportion to theirdisk bandwidth entitlements.

diskless cluster serverA system that supports disk activity for diskless client nodes.

diskless file system bufferA buffer pool that is used only by the diskless server for disklesscluster traffic.

dispatcherA module of the kernel responsible for allocating CPU resources amongseveral competing processes.

DSI log fileA log file, created by OV Performance Server Agent’s DSI (data sourcreintegration) programs, that contains self-describing data..

empty spaceThe difference between the maximum size of a log file and its currentsize.


547

error (LAN)Unsuccessful transmission of a packet over a local area network (LAN).Inbound errors are typically checksum errors. Outbound errors aretypically local hardware problems.

exec fill pageWhen a process is 'execed' the working segments of the process aremarked as copy on write. Only when segments change are they copiedinto a separate segment private to the process that is modifying thepage.

extract programThe OV Performance Server Agent’s program that allows you to extractdata from raw or previously extracted log files, summarize it, and writeit to extracted log files. It also lets you export data for use by analysisprograms and other tools.

extracted log fileAn OV Performance Server Agent log file containing a user-definedsubset of data extracted (copied) from raw or previously extracted logfile. It is formatted for optimal access by OV Performance Manager.Extracted log files are also used for archiving performance data.

file IOIO activity to a physical disk. It includes file system IOs, system IOs tomanage the file system, both raw and block activity, and excludesvirtual memory management IOs.

file lockA file lock guarantees exclusive access to an entire file, or parts of afile.

file systemThe organization and placement of files and directories on a hard disk.The file system includes the operating system software's facilities fornaming the files and controlling access to these files.

file system activityAccess calls (read, write, control) of file system block IO files containedon disk.

file system swapFile system space identified as available to be used as swap. This is alower performance method of swapping as its operations are processedthrough the file system.


548

file tableThe table contains inode descriptors used by the user file descriptorsfor all open files. It is set to the maximum number of files the systemcan have open at any one time.

forkA system call that enables a process to duplicate itself into twoidentical processes - a parent and a child process. Unlike the vforksystem call, the child process produced does not have access to theparent process' memory and control.

free listThe system keeps a list of free pages on the system. Free list points toall the pages that are marked free.

free memoryMemory not currently allocated to any user process or to the kernel.

GlancePlusAn online diagnostic tool that displays current performance datadirectly to a user terminal or workstation. It is designed to assit you inidentifying and troubleshooting system performance problems as theyoccur.

globalA qualifier implying the whole system. Thus "global metrics" aremetrics that describe the activities and states of each system.Similarly, application metrics describe application activity; processmetrics describe process activity.

global log fileThe raw log file, logglob, where the scopeNT collector placessummarized measurements of the system-wide workload.

idleThe state in which the CPU is idle when it is waiting for the dispatcher(scheduler) to provide processes to execute.

idle biodThe number of inactive NFS daemons on a client.


549

initial groupThe first Process Resource Manager (PRM) group listed in a PRM userrecord of the PRM configuration file. Any processes a user initiates aregenerally run in the user's initial group.

inodeA reference pointer to a file. This reference pointer contains adescription of the disk layout of the file data and other information,such as the file owner, access permissions, and access times. Inode is acontraction of the term 'index node'.

inode cacheAn in memory table containing up-to-date information on the state of acurrently referenced file.

interesting processA filter mechanism that allows the user to limit the number of processentries to view. A process becomes interesting when it is first created,when it ends, and when it exceeds user-defined thresholds for CPU use,disk use, response time, and so on.

interruptHigh priority interruptions of the CPU to notify it that something hashappened. For example, a disk IO completion is an interrupt.

intervalsSpecific time periods during which performance data is gathered.

ioctlA system call that provides an interface to allow processes to control IOor pseudo devices.

IO doneThe Virtual Memory Management (VMM) system reads and writesfrom the disk and keeps track of how many IOs are completed by thesystem. Since IOs are asynchronous, they are not completedimmediately.

Sometimes IOs done can be higher than IO starts, since some of theIOs that are started in the previous interval can be completed.

IO startThe Virtual Memory Management (VMM) system reads and writesfrom the disk and keeps track of how many IOs are started by thesystem. Since IOs are async, they are not completed immediately.


550

InterProcess Communication (IPC)Communication protocols used between processes.

kernelThe core of the UNIX operating system. It is the code responsible formanaging the computer's resources and performing functions such asallocating memory. The kernel also performs administrative functionsrequired for overall system performance.

kernel tableAn internal system table such as the Process Table or Text Table. Atable's configured size can affect system behavior.

last measurement resetWhen you run a performance product, it starts collecting performancedata. Cumulative metrics begin to accumulate at this time. When youreset measurement to zero, all cumulative metrics are set to zero andaverages are reset so their values are calculated beginning with thenext interval.

load averageA measure of the CPU load on the system. The load average is definedas an average of the number of processes running and ready to run, assampled over the previous one-minute interval of system operation.The kernel maintains this data.

lock missThe Virtual Memory Management (VMM) system locks pages forsynchronization purposes. If the lock has to be broken for any reasonthat is considered a lock miss. Usually this is a very small number.

logappl (application log file)The raw log file that contains summary measurements of processes ineach user-defined application.

logdev (device log file)The raw log file that contains measurements of individual device (disk,logical volume, network interface) performance.

logglob (global log file)The raw log file that contains measurements of the system-wide, orglobal, workload.


551

logindexThe raw log file that contains information required for accessing datain the other log files.

logproc (process log file)The raw log file that contains measurements of selected interestingprocesses.

logtran (transaction log file)The raw log file that contains measurements of transaction trackingdata.

log filesPerformance measurement files that contain either raw or extractedlog file data.

logical IOA read or write system call to a file system to obtain data. Because ofthe effects of buffer caching, this operation may not require a physicalaccess to the disk if the buffer is located in the buffer cache.

macroA group of instructions that you can combine into a single instructionfor the application to execute.

major faultA page fault requiring an access to disk to retrieve the page.

measurement interfaceA set of proprietary library calls used by the performance applicationsto obtain performance data.

mem entitlementThe percentage of memory guaranteed to a particular Process ResourceManager (PRM) group when the total system memory use is at itsmaximum. The system administrator assigns the memory entitlementfor each PRM group in a PRM configuration file. The minimumentitlement for groups other than the system group is 1%. PRMdistributes unused time to other groups in proportion to their memoryentitlements.

memory pressureA situation that occurs when processes are requesting more memoryspace than is available.


552

memory upperboundThe upper memory threshold is a flexible (soft) upper boundary. If agroup's memory use is above its upper memory threshold and systemmemory use is approaching 100%, then regardless of whether othergroups are currently in need of memory, Process Resource Manager(PRM) will control the group's memory use by suppressing the group'sprocesses.

memory swap spaceThe part of physical memory allocated for swapping.

memory thrashingSee thrashing.

message buffer poolA cache used to store all used message queue buffers on the system.

message queueThe messaging mechanism allows processes to send formatted datastreams to arbitrary processes. A message queue holds the buffersfrom which processes read the data.

message tableA table that shows the maximum number of message queues allowedfor the system.

metricA specific measurement that defines performance.characteristics.

midaemonA process that monitors system performance and creates counters fromsystem event traces that are read and displayed by performanceapplications.

minor faultA page fault that is satisfied by a memory access (the page was not yetreleased from memory).

mount/unmountThe process of adding or removing additional, functionally-independentfile systems to or from the pool of available file systems.


553

Network Node Manager (NNM)A network management application that provides the network mapused by OV Performance Manager.

network timeThe amount of time required for a particular network request to becompleted.

NFS callA physical Network File System (NFS) operation a system has receivedor processed.

NFS clientA node that requests data or services from other nodes on the network.

NFS IOA system count of the NFS calls.

NFS Logical IOA logical I/O request made to an NFS mounted file system.

NFS-mountedA file system connected by software to one system but physicallyresiding on another system's disk.

NFS serverA node that provides data or services to other nodes on the network.

NFS transferTransfer of data packets across a local area network (LAN) to supportNetwork File System (NFS) services.

niceAltering the priority of a time-share process, using either thenice/renice command or the nice system call. High nice values lessenthe priority; low nice values increase the priority.

nodeA computing resource on a network, such as a networked computersystem, hub, or bridge.


554

normal CPUCPU time spent processing user applications which have not been real-time dispatched or niced.

outbound read/writeThe designation used when a local process requests a read from orwrite to a remote system via NFS.

o/f (overflow)This designates that the measurement software has detected a numberthat is too large to fit in the available space.

OV Performance ManagerA tool that provides integrated performance management for multi-vendor distributed networks. Uses a single workstation to monitorenvironment performance on networks that range in size from tens tothousands of nodes.

packetA unit of information that is transferred between a server and a clientover the LAN.

packet in/outA request sent to the server by a client is an "in" packet. A requestsent to a client by the server is an "out" packet.

pageA basic unit of memory. A process is accessed in pages (demandpaging) during execution.

pagedaemonA system daemon responsible for writing parts of a process' addressspace to secondary storage (disk) to support the paging capability of thevirtual memory system.

page faultAn event recorded when a process tries to execute code instructions orto reference a data page not resident in a process' mapped physicalmemory. The system must page-in the missing code or data to allowexecution to continue.

page freedWhen a paging daemon puts a page in the free list, it is considered aspage freed.


555

page in/page outMoving pages of data from virtual memory (disk) to physical memory(page in) or vice versa (page out).

page reclaimVirtual address space is partitioned into segments, which are thenpartitioned into fixed size units called pages. There are usually twokinds of segments: persistent segments, and working segments.

Files containing data or executable programs are mapped intopersistent segments. A persistent segment (text) has a permanentstorage location on disk so the Virtual Memory Manager writes thepage back to that location when the page has been modified and it is nolonger kept in real memory. If the page has not changed, its frame issimply reclaimed.

page requestA page fault that has to be satisfied by accessing virtual memory.

page scanThe clock hand algorithm used to control page and to select pages forremoval from system memory. It scans pages to select pages forpossible removal.

page spaceThe area of a disk or memory reserved for paging out portions ofprocesses or swapping out entire processes. Also known as swap space.

page stealOccurs when a page used by a process is taken away by the VirtualMemory Management system.

pagein routineA kernel routine that brings pages of a process' address space intophysical memory.

pageout routineA kernel routing that executes when physical memory space is scarce,and the pagedaemon is activated to remove the least-needed pagesfrom memory by writing them to swap space or to the file system.

parm fileThe file containing the parameters used by OV Performance ServerAgent’s scopeNT data collector to customize data collection. Also usedto define your applications.


556

perflbd.rcThe configuration file that contains entries for one or more datasources, each of which represents a scopeNT or DSI data source. Seealso repository server.

performance distribution rangeAn amount of time that you define with the range= keyword in thetransaction tracking configuration file, ttd.conf.

perfstatThe script used for viewing the status of all Hewlett-Packardperformance products on your system.

pfaultsMost resolvable pfaults (protection faults) are caused by copy on writes(for example, writing to private memory segments). Most other pfaultsare protection violations (for example, writing to a read-only region)and result in SIGBUS. See mprotect(2).

physical IOA input/output operation where data is transferred from memory todisk or vice versa. Physical IO includes file system IO, raw IO, systemIO, and virtual memory IO.

physical memoryThe actual hardware memory components contained within yourcomputer system.

PIDA process identifier - a process' unique identification number thatdistinguishes it from all other processes on the system. PPID is aparent process identifier - the process identifier of a process that forkedor vforked another process.

pipeA mechanism that allows a stream of data to be passed between readand write processes.

priorityThe number assigned to a PID that determines its importance to theCPU scheduler.


557

PRM configuration fileThe Process Resource Manager (PRM) configuration file defines PRMgroups, CPU entitlements and caps, memory entitlements and caps,user access permissions, application/PRM group associations, and diskbandwidth entitlements. The default PRM configuration file is/etc/prmconf.

The configuration file can contain five types of records; however, you donot have to use each type of record. The record types are:

• Group (required) - defines PRM groups and CPU entitlements

• Memory - defines real memory entitlements and caps

• User - specifies which PRM groups a user can access

• Application - defines associations between applications and PRM groups

• Disk - defines disk bandwidth entitlements for a specific logical volume group

PRM groupA group of users and/or applications that is entitled to a minimumpercentage of CPU, and optionally memory or disk bandwidth. ProcessResource Manager (PRM) groups are defined in the PRM configurationfile /etc/prmconf or a configuration file you specify. Each PRM grouphas a name, a number (PRMID), and a CPU entitlement. Groups canoptionally have memory entitlements and disk bandwidthentitlements.

PRMIDAn integer between 0 and 47, inclusive, that uniquely identifies aProcess Resource Manager (PRM) group in the PRM configuration file.PRMID 0 is reserved for the System Group. PRMID 1 is reserved forthe User Default Group.

proc tableThe process table that holds information for every process on thesystem.

processThe execution of a program file. This execution can represent aninteractive user (processes running at normal, nice, or real-timepriorities) or an operating system process.

process blockA process block occurs when a process is not executing because it iswaiting for a resource or IO completion.


558

process deactivation/reactivationA technique used for memory management. Process deactivationmarks pages of memory within a process as available for use by othermore active processes. A process becomes a candidate for deactivationwhen physical memory becomes scarce or when a system startsthrashing.

Processes are reactivated when they become ready to run.

process stateDifferent types of tasks executed by a CPU on behalf of a process. Forexample: user, nice, system and interrupt.

pseudo terminal (pty)A software device that operates in pairs. Output directed to onemember of the pair is sent to the input of the other member. Input issent to the upstream module.

queueA waiting line in which unsatisfied requests are placed until a resourcebecomes available.

raw IOUnbuffered input/output that transfers data directly between a diskdevice and the user program requesting the data. It bypasses the filesystem's buffer cache. Also known as character mode. Compare withblock mode.

raw log fileAn OVPerformance Server Agent file into which scopeNT logs collecteddata. It contains summarized measurements of system data. Seelogglob, logappl, logproc, logdev, logtran, and logindx.

read byte rateThe rate of kilobytes per second the system sent or received doing readoperations.

read rateThe number of NFS and local read operations per second a system hasprocessed. Read operations consist of getattr, lookup, readlink,readdir, null, root, statfs, and read.

Read/write QlenThe number of pending NFS operations.


559

read/write system callA request that a program uses to tell the kernel to perform a specificservice on the program's behalf. When the user requests a read, a readsystem call is activated. When the user requests a write, a writesystem call is activated.

real timeThe actual time in which an event takes place.

real time cpuTime the CPU spent executing processes that have a real-time priority.

remote swappingSwapping that uses swap space from a pool located on a differentsystem's swap device. This type of swapping is often used by disklesssystems that swap on a server machine.

repeat timeAn action that can be selelcted for performance alarms. Repeat timedesignates the amount of time that must pass before an activated andcontinuing alarm condition triggers another alarm signal.

repository serverA server that provides data to the alarm generator and the OVPerformance Manager analysis product. There is one repository foreach data source configured in the perflbd.rc configuration file. Adefault repository server, provided at start up, contains a single datasource consisting of a scopeNT log file set.

reserved swap spaceArea set aside on your disk for virtual memory.

resident bufferData stored in physical memory.

resident memoryInformation currently loaded into memory for the execution of aprocess.

resident set sizeThe amount of physical memory a process is using. It includes memoryallocated for the process' data, stack, and text segments.


560

resizeChanging the overall size of a raw log file.

response timeThe time spent to service all NFS operations.

roll backDeleting one or more days worth of data from a raw log file with theoldest data deleted first. Roll backs are performed when a raw log fileexceeds its maximum size parameted,

rxlogThe default extract log file created when data is extracted from raw logfiles.

SCOPEThe OV Performance Agent default data source that contains a scopeuxor scopeNT global log file set.

scopeuxThe MeasureWareServer Agent data collector program that collectsperformance data and writes (logs) it to raw log files for later analysisor archiving.

scopeNTThe MeasureWareServer Agent data collector program that collectsperformance data and writes (logs) it to raw log files for later analysisor archiving.

scopeux log filesThe raw log files that are created by the scopeux collector: logglob,logappl, logproc, logdev, logtran, and logindx.

scopeNT log filesThe raw log files that are created by the scopeNT collector: logglob,logappl, logproc, logdev, logtran, and logindx.

semaphoreSpecial types of flags used for signaling between two cooperatingprocesses. They are typically used to guard critical sections of codethat modify shared data structures.


561

semaphore tableMaximum number of semaphores currently allowed for the system.

service level agreementA document prepared for a business critical application that explicitlydefines the service level objectives that IT (Information Technology) isexpected to deliver to users. It specifies what the users can expect interms of system response, quantities of work, and system availability.

service level objectiveA definable level of responsiveness for a transaction. For example, ifyou decide that all database updates must occur within 2 seconds, setthe Service Level Objective (SLO) for that transaction as slo=2.

shared memorySystem memory allocated for sharing data among processes. It includesshared text, data and stack.

shared memory poolThe cache in which shared memory segments are stored.

shared memory segmentA portion of a system's memory dedicated to sharing data for severalprocesses.

shared memory tableA list of entries that identifies shared memory segments currentlyallocated on your system.

shared text segmentCode shared between several processes.

signalA software event to notify a process of a change. Similar to a hardwareinterrupt.

sleeping processA process that either has blocked itself or that has been blocked, and isplaced in a waiting state.

socket operationA process that creates an endpoint for communication and returns adescriptor for use in all subsequent socket-related system calls.


562

start of collectionWhen you run a performance product, it starts collecting performancedata.

summary dataThe time period represented in one data point of a performancemeasurement. Summary levels can be five minutes, one hour, and oneday.

swapA memory management technique used to shuttle information betweenthe main memory and a dedicated area on a disk (swap space).Swapping allows the system to run more processes than couldotherwise fit into the main memory at a given time.

swap in/outMoving information between the main memory and a dedicated(reserved) area on a disk. ''Swapping in'' is reading in to virtualmemory; ''swapping out'' is reading out from virtual memory.

swap spaceThe area of a disk or memory reserved for swapping out entireprocesses or paging out portions of processes. Also known as pagespace.

system callA command that a program uses to tell the kernel to perform a specificservice on the program's behalf. This is the user's and applicationprogrammer's interface to the UNIX kernel.

system codeKernel code that is executed through system calls.

system CPUTime that the CPU was busy executing kernel code. Also called kernelmode.

system diskPhysical disk IO generated for file system management. These includeinode access, super block access and cylinder group access.

system groupThe PRM group with PRMID 0. PRM places all system processes, suchas init and swapper, in this group by default.


563

system interrupt handling codeKernel code that processes interrupts.

terminal transactionA terminal transaction occurs whenever a read is completed to aterminal device or MPE message file. On a terminal device, a read isnormally completed when the user presses the return or the enter key.Some devices such as serial printers may satisfy terminal reads byreturning hardware status information. Several metrics are collectedto characterize terminal transactions.

The FIRST_RESPONSE_TIME metric measures the time between thecompletion of the read and the completion of the first write back to thatdevice. This metric is most often quoted in bench marks as it yields thequickest response time. For transactions which return a large amountof data to the terminal, such as reading an electronic mail message, thetime to first response may be the best indicator of overall systemresponsiveness.

The RESPONSE_TIME_TO_PROMPT metric measures the timebetween the completion of the read and the posting of the next read. Itis the amount of time that a user must wait before being able to enterthe next transaction. This response time includes the amount of timeit took to write data back to the terminal as a result of the transaction.The response time to prompt is the best metric for determining thelimits of transaction throughput.

The THINK_TIME metric measures the time between posting a readand its completion. It is a measure of how much time the user took toexamine the results of the transaction and then complete entering thenext transaction. Transaction metrics are expressed as average timesper transaction and as total times in seconds. Total times arecalculated by multiplying the average time per transaction times thenumber of transactions completed.

Terminal transactions can be created by interactive or batch processesthat do reads to terminal devices or message files. Reads to terminaldevices or message files done by system processes will not be countedas transactions.

text segmentA memory segment that holds executable program code.

thrashingA condition in which a system is spending too much time swappingdata in and out, and too little time doing useful work. This ischaracteristic of situations in which either too many page faults arebeing created or too much swapping is occurring. Thrashing causes thesystem's performance to degrade and the response time for theinteractive users to increase.


564

threadpool queueA queue of requests waiting for an available server thread.

thresholdNumerical values that can be set to define alarm conditions. When athreshold is surpassed, an alarm is triggered.

tooltipDisplay of the full text of a truncated data string in a row-columnformated GlancePlus report window. Tooltips are enabled and disabledby choosing Tooltips from the window's Configure menu or byclicking the "T" button in the upper right corner of the window.

transactionSome amount of work performed by a computer system on behalf of auser. The boundaries of this work are defined by the user.

transaction trackingThe technology used by OV Performance Agent and GlancePlus thatlets information technology (IT) managers measure end-to-endresponse time of business application transactions.

trapSoftware interrupt that requires service from a trap handler routine.An example would be a floating point exception on a system that doesnot have floating point hardware support. This requires the floatingpoint operations to be emulated in the software trap handler code.

trap handler codeTraps are measured when the kernel executes the code in the traphandler routine. For a list of trap types, refer to the file/usr/include/machine/trap.h.

ttd.confThe transaction tracking configuration file where you define eachtransaction and the information to be tracked for each transaction,such as transaction name, performance distribution range, and servicelevel objective.

unmount/mountThe process of removing or adding functionally-independent filesystems from or to the root file system.


565

update intervalThe interval of time between updates of the metrics that display in areport window or graph.

user codeCode that does not perform system calls.

user CPUTime that the CPU was busy executing user code. This includes timespent executing non-kernel code by daemon processes. It does notinclude CPU time spent executing system calls, context switching, orinterrupt handling.

user default groupThe PRM group with PRMID 1. PRM uses this group as the initialgroup for any user who does not have a PRM user record in thecurrently loaded PRM configuration file.

user diskPhysical disk IO generated by accessing the file system.

utility programAn OVPerformance Server Agent program that lets you check parm fileand alarmdef file syntax, resize log files, scan log files for information,and obtain alarm information from historical log file data.

vfault CPUCPU time spent handling page faults.

vfaultsA vfault (virtual fault) is the mechanism that causes paging. Accessingan unmapped valid page causes a resolvable vfault. Accessing anillegal address results in a SIGSEGV.

vforkA version of the fork system call that spawns a child process that iscapable of sharing code and data with its parent process.

virtual memorySecondary memory that exists on a portion of a disk or other storagedevice. It is used as an extension of the primary physical memory.


566

virtual memory IOThe virtual memory reads or writes from the disk for memory mappedfiles, and for paging out pages from paging area (swap area). Since allthe files are memory mapped, all the reads or writes are virtualmemory reads or writes as well. The computational memory of theprocesses that are changing is paged out if necessary to the swap areaand read or written from there again.

write byte rateThe rate of kilobytes per second the system sent or received duringwrite operations.

write rateThe number of NFS and local write operations the local machine hasprocessed per second. Write operations include setattr, writecache,create, remove, rename, link, symlink, mkdir, rmdir, and write.

X-AxisThe horizontal scale on a graph.

Y-AxisThe vertical scale on a graph.

zero fill pageWhen pages are requested by the processes they are usually allocatedby the Virtual Memory Management system and filled with zeros.

Documents

HP OpenView GlancePlus · Hewlett-Packard makes no warranty of any kind with regard to this material, including but not limited to, the implied warranties of merchantability and fitness