Failover validation for ServiceGuard cluster

[timus]

Hello,

I need to validate the existing cluster configuration of our environement (HP-UX 11.23/11.31 with ServiceGuard)

Is there a document or checklist for Service Guard failover validation without impacting the production environement?

I don't want to do actual failover testing, but only want to make sure that the existing clusters are properly configured and the applications will failover to the designated failover node when needed.

would appreciate if someone can give me a step by step procedure to validate failover configurations.

[Jeff_Traigle]

Without actually doing a failover, you can't really validate that everything will work as expected. Your package control scripts could have bugs in them that could be overlooked by visual inspection.

You can use "cmviewcl -v" to validate you have certain things setup as you want (e.g. cluster lock disks, LAN interfaces, policy parameters, node switching settings). You can use cmgetconf to visually validate some other items (e.g. timeout values).

[Steven E. Protter]

Shalom,

You must test fail over. No alternative.

Many shops have a lab environment to test these.

Basic tests:

1) Pull the main network connection(s) on one node. All services should fail over to the other node.
2) TOC (button on the back) one node. All services should fail over.
3) Kill a process monitored by your SG package. Process should either restart or services should fail over.

SEP

[Terry.giblin]

Dear Timus,

That's easy.

20 years experience remembering to,

1. Check your tape backups work.

2. Check your system backups work.

3. Check your disaster recovery plan works.

4. Check your Cluster configuration works.

5. Check your SAN configuration works.

6. Check your Network configuration works.

7. Check your Database configuration works.

What could be simpler.

Everything works in theory, but theory also has to withstand any real life test.

Terry Giblin

[Terry.giblin]

Alternatively, you can plan and arrange an agreed scheduled downtime or outgage or a maintenance window.

Your company save's time and money.

Or just wait a little while longer and I guarantee, an opportunity will present itself.

Normally when you least expect it.

[Nesan]

You can follow this manual

http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c02056246/c02056246.pdf

Thank you
Regards
NS