13
Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean- Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky CM26 Mar 26, 2010

Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

  • View
    216

  • Download
    2

Embed Size (px)

Citation preview

Page 1: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

Computing Panel Discussion Continued

Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick

Hanlet, Henry NebrenskyCM26 Mar 26, 2010

Page 2: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

What We Need to Do• MICE needs to run the DAQ to record data, monitor the

equipment, and control the various components of the experiment. We need to monitor the data-taking and reconstruct live data to ensure data quality. We need to archive actual running parameters. We need a Database to automatically keep track of run conditions which then feeds into the analysis software. The analysis software needs to be able to reconstruct the data and run simulations of the experiment.

• In order to facilitate these requirements, MICE needs network access both at RAL and from outside of RAL.

Page 3: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

What We Need to Do2

• At RAL:– We need access to off-site computers and the ability to copy

files to those computers.– We need access to the web, web email, and other email

sources.– We need to be able to work on other computers as if we were

there in person (VPN).– We need a system to maintain code development: CVS for

analysis, Bazaar for C&M.– We need to allow MICE at RAL to read and write to the eLOG.– We need access to the Database (through an interface – API)

for analysis. The interaction between the user and interface should remain transparent even if the interface needs to modify how it interacts with the Database.

Page 4: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

What We Need to Do3

• From outside of RAL:– We need to allow expert access to MICE machines in the control room

and on the PPD network. – They must be able to access the data in real-time to debug.– Some experts must be able to remotely display a terminal from the

MLCR to a remote computer to make changes. Others need to remotely display information for monitoring purposes. (ex. EPICS, DS cooldown, Target calibration data)

– We need offsite experts to be able to see web based information like the webcams and CKOV temperature/humidity monitor.

– External MICE must be able to read/write to the eLOG.– External MICE also need access to the Database.

• Access details:– PPD Network CAN see MiceNet– Visitors Network NOT see MiceNet– Outside of RAL NOT see MiceNet directly– Federal ID into RAL as if on PPD (ex.)

Page 5: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

How We Do All of This• Hardware

– Machines in MLCR, Lab7 (tracker), R76 (MICE Office), R78 (Target test area), Visitor laptops, Analysis Machine/Farm?

– SSH Bastion– Web Services Machine– Database Machine

• Networks Used– MiceNet – control room network– PPD – RAL particle physics dept. network– Visitor’s Network (Guest)– Networks available in MLCR – PPD, ISIS

• Access details:– PPD Network CAN see MiceNet– Visitors Network NOT see MiceNet– Outside of RAL NOT see MiceNet directly– Federal ID into RAL as if on PPD (ex.)

Page 6: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

Henry Nebrensky – CM26 – 24 March 2010

External Connectivity MICO Slide

SSH

EPICS Gateway

DB API

DB

Outside World Micenet / MLCR

DB API

eLog

SSH Bastion

Config Database

“Web” services

EPICS archiver web interface

ssh

SSH / web services

EPICS

Spare node

Grid clientsGrid Transfer Box

6

PPD-Grid managed

SSH+ analysis

code

Analysis Node/Farm ssh

MICE managed

PPD-IT supervised

Page 7: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

What We Are Doing• Replicating Functionality of heplnw17 - three new servers already in R11. MICE SSH Bastion

– Provides access into RAL site– Once in, as if on PPD network– Then can get to MiceNet

• No direct access to MiceNet– Copy files out (ie. Data)

• Note: ask Mike about SCP out?– Bring through xterm windows to outside (x forwarding)– ON PPD NETWORK– Will provide all access requested so far…

2. Web Services Machine– eLOG (if still want it at RAL)

• ON PPD NETWORK Least downtime possible• MICE have direct access• Global MICE read/write access

– User interface for Database – Stuff related to forwarding web related traffic from the control room to IIT server (like

webcams)

3. Database Machine• ON PPD NETWORK

Page 8: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

What We Are Doing2

• SSH Bastion– Malcolm has kindly agreed to pick this up (with written approval by

management)– Will speak to RAL PPD person Tuesday to push through firewall ports – IDS meeting at Fermilab – then back to finish eta. Beginning May

• Move Windows machine out of MLCR into R76– Not needed in control room– User machine on wired Visitors Network

• Provide Analysis Machine/Farm at RAL?– Must be on PPD network to use GRID

• Federal IDs for MICE– Longer term process– Longer term access (SSH Bastion provides easy way to give new person

access if needed on short notice)– Should solve new Guest access issues

• ≤1 month now

Page 9: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

Computing Support

• Who is responsible for each machine/system?

• The 3 new servers in R1 are on the PPD Network

• Requesting official PPD support for these machines – MICE SSH Bastion, Web Services Machine,

Database Machine

Page 10: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

Computing Support• Who is responsible for each machine/system?• MLCR Machines

– C&M = James Leaver - GRID = Henry– DAQ = Jean-Sebastien Graulich– OnRec = Linda Coney - Target =? James– Linde Computer = Linde– Willie’s Laptop = ?Craig, Malcolm?– Webcams & Printer = Craig

• Other– MOM Laptop = Malcolm– Ash computer = PPD– Lab 7 = FNAL & James & Geneva?– Visiting Laptops = Visitors

• Backups…

Page 11: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

Roles to Define

• Database Administration• eLOG Management• Web cams - ?Craig• MICE Network Administrator/liason – Mike Courthold, Craig

Macwaters (deputy), Henry Nebrensky• Backup Management• GRID

– Data Transfer – Henry Nebrensky– Software Management – Vassil Verguilov– Data Archivist - Henry Nebrensky

• Repository Management• MICE software updates (EPICS, G4MICE, etc)• Simulation Production• Real Data Production• Analysis Production

Page 12: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky

Questions for You

• What are requirements for future systems? – Spectrometer Solenoid

• How controls/monitoring handled?• What expert access will be needed?

– LH2• Is this handled by RAL?

– RF• Same questions as Spectrometer Solenoid

DL

DLDL

Page 13: Computing Panel Discussion Continued Marco Apollonio, Linda Coney, Mike Courthold, Malcolm Ellis, Jean-Sebastien Graulich, Pierrick Hanlet, Henry Nebrensky