upgrading OS in cluster env

[himacs]

i have 2 node cluster(active-passive).Now i have to upgrade the OS from 11.23 to 11.31.Please tell me the steps to follow.

Since in 11.31 FS naming convention is different from 11.23, we have to regenerate cluster config file...?

cluster version is 11.17
Please suggest.

himacs

[sangilak]

Hi Himacs,


Your best source will be the "Managing ServiceGuard Seventeenth" guide:

http://bizsupport1.austin.hp.com/bc/docs/support/SupportManual/c02057890/c02057890.pdf

The section on software updates start on page 413 and there is information on the 5 possible types of upgrades:

â ¢ rolling upgrade
â ¢ rolling upgrade using DRD
â ¢ non-rolling upgrade
â ¢ non-rolling upgrade using DRD
â ¢ migration with cold install

Although I have not done such an upgrade myself, it appears that most go for either a rolling upgrade using Ignite-UX or a migration with cold install.


Regards,

sangilak

[Rita C Workman]

Your running an active/passive cluster.
You don't mention what kind of downtime you can have.
So here's a suggestion.

Prior to updating O/S on each server:
1. Make solid backups
2. Make recovery backups.
3. Document your system

Remove your 'passive' node from your existing cluster, leaving you a single node cluster.

Update-UX the 'passive' node to 11.31 and confirm all is solid.
Take this 'passive' node and create a brand new single node cluster.

When you can get downtime for your package(s) on the running old single-node server, you will remove each package from the "now-old-cluster" and add the package into the "now-new-cluster".
Once you have all the packages up and running on the new single node cluster, you can then do the Update-UX the 'old cluster' to bring it up to 11.31.

Then check this server to ensure everything looks good afterwards, and then add it to the "now-new-cluster".

You have now updated on both sides of your cluster with an active/passive setup again.
And the downtime was minimal with just the little bit of time it takes to move your package(s) to the new cluster and start them back up.

Make fresh backups, fresh recovery backups and you are done!

Regards,
Rita