Hardware monitoring with HPSG

[E. Louvet]

Hello,
I'm encountering non sense behaviour with monitoring packages under RHEL 5.1 using HP SG 11.18.0.4.

First I have configured a package with the module monitored_subnet.

If the ethernet link goes done, the package fails as expected.

But at startup, if the link is down, the package starts and stay up, so the problem is not seen at all.

I encounter the same behaviour with a package starting cmresserviced monitoring a disk LUN based on a MSA 1000.

if the FC link is broken, the package starts well and fail only when (I guess) cmresserviced check the lun a few seconds later. I was expecting the package to not start at all.

These 2 points could be corrected with a customisation of the operations_sequence, in inserting a script causing the package not to start if a resource is down, but the comments in the package skeleton says it would brake warranties and support.

And here is the last problem.

This is a deployment thing, I have the need of build an image of a cluster to be deployed in many different sites. So I'm bored with the MAC address stored in the cmclconfig file.

The cmclconfig file could be copied in another identical cluster and starts well neverless all MAC addresses have changed, but the HP documentation recommends to re-configure the cluster after changing a Lan card.

What are the risks ? limitations ?


Regards.

[E. Louvet]

Hum ... Ok, no one is monitoring hardware under serviceguard ?

Is HPSG a Wrong Product for HA ?