Automated HACMP Upgrade

Embed Size (px)

Citation preview

  • 8/13/2019 Automated HACMP Upgrade

    1/3

    HACMP Upgrade Script

    There are 3 scripts involve to successfully upgrade HACMP cluster. These scripts arelocated on all NIM servers under /nim/MaintLevels/HACMP/CONFIGURE directory, so

    the very first thing that you need to do is to nfs mount /nim/MaintLevels directory to theHACMP nodes.

    NOTE:Make sure you nfs mount /nim/MaintLevels directory and NOT/nim/MaintLevels/HACMP/CONFIGURE directory. The scripts will get some of its data

    under /nim/MaintLevels. If you did not nfs mount the directory correctly, the script will

    fail.

    PLEASE READ:

    1. HACMP Must be stop on both nodes before performing all the steps below.

    2. Go to our website mossweb.metlife.com, select configuration then HACMP

    Information, then find the nodename from the list and click on the cluster name

    link. This will open a pop-up window containing the HACMP configuration. You

    will need this info when performing the steps below.

    3. In Phase 2: Setup Network, when ask for BOOT IP, enter the 192.168 ip address

    (standby ip address from old config), and when ask for Persistent IP add, enter the

    IP address (boot ip address from old config).

    PHASE 0: nfs mount nimserver /nim/MaintLevels on both nodes

    PHASE 1: Upgrade HA Code - To be perform on both nodes

    1. Execute:

    /mnt/HACMP/CONFIGURE/upgrade_hapfor Primary Node

    /mnt/HACMP/CONFIGURE/upgrade_hasfor Secondary Node

    NOTE:For active-active node, you have to choose 1 node as primary and the

    other node as secondary.

    This will prompt you if this is HADPAR or not, so make sure you know if this isHADPAR or not.

    The script does the following:o gather some basic information needed to configure HA later, this config

    will be saved in /usr/local/HACMP directory.

  • 8/13/2019 Automated HACMP Upgrade

    2/3

    o run an HACMP snapshot and save it in /usr/local/HACMP directory.o Remove old HA code and install new HA codeo Set the high/low watermark and syncd valueso update root .profileo create netmon.cf fileo

    copy cluster_ipconfig script to /usr/local/HACMP directory.

    2. unmount /mnt

    You can run this script on both nodes at the same time.

    PHASE2: Set up Network ==> to be perform on both nodes

    1. Login directly to the Console (not telnet or ssh)

    2. (make sure there is no NFS mounted before doing this step) Remove TCP/IP config for

    en interface, (ifconfig down, ifconfig detach, rmdev, flush route.....etc). (don't touch the

    TSM interface).

    3. Configure Etherchannel (If any) This is required for those partitions that dont use

    virtual ethernet. Also, please make sure to run cfgmgr first after removing en# and ent#devices from step 2. Make sure also to check the adapter type and the speed settings (if

    10/100mbps card, set it to 100full, if 10/100/1000mbps card, leave it to auto_neg).

    Note also that when creating etherchannel (smitty etherchannel), make sure to select one

    adapter as primary, hit enter, then go to the backup adapter and press F4 (or esc 4) toselect the backup adapter.

    NOTE:You don't need to setup tcpip after configuring the etherchannel, step #4below will perform that for you.

    4. run /usr/local/HACMP/cluster_ipconfig

    You will be prompted for the network information. You will also be prompted to

    enter the interface name, please enter the etherchannel interface name (en#).The script does the following;

    configure tcpip update /etc/hosts

    5. Once done, please check/compare the new /etc/hosts file with /etc/hosts.migrateha, and

    add anything you think important alias names, and/or host entries.

  • 8/13/2019 Automated HACMP Upgrade

    3/3

    Perform this step on secondary node, you can actually perform this step on both

    server at the same time (just make sure you don't get confused :-)).

    PHASE3: Configure HA ==> You only need to run this on the PRIMARY

    NODE: (Select any node if active-active).

    1. nfs mount /nim/MaintLevels

    2. cd /usr/local/HACMP

    3. Edit haadapter.cfg file, and replace the interface name/value for all INTERFACE entryto the etherchannel interface name.

    4. run /mnt/HACMP/CONFIGURE/configure_ha

    The script performs HA configuration similar to the NEW HACMP setup script, it

    will pick up the following information that was gathered in PHASE1:

    type of heartbeat application startup/stop script name etc....

    Then convert them to our standard naming convention.

    5. Once done, copy the start/stop script from primary node to the secondary node. You

    can use any method of transfer you want (ftp or sftp or scp), just make sure that the

    scripts have execute permission.

    6. unmount /mnt

    7. run HA verification and synchronization

    8. start HA and start testing...

    END OF DOCUMENT