Is there a disk monitor in SG to monitor my xp512 disk

[Dan Hoyt]

I'm fairly new to SG and am setting up a simple 2 node cluster. I'm testing failover now and noticed when I unplug my single fibre connection to my XP512, I lose connection to my disk but the box doesn't failover.

Can someone point me in the correct direction on some kind of monitor that will talk to SG and TOC my box for failover?

Thanks,
Dan

[Stephen Doud · Accepted solution]

Document UMCSGKBRC00012483 discusses how to create a High Availability Monitor that will trigger Serviceguard to fail the package to the adoptive node.
The caveat is that the node where the connection is lost must TOC (reboot) in order to force the failover because HP-UX has no mechanism to forcibly umount a file system when I/O transactions are pending.

The RESOURCE_NAME needed is /vg/vgName/pv_summary

See this document for more details:
http://docs.hp.com/en/B5735-90001/index.html

[Simon Hargrave]

On a sidenote, given that you are building a cluster for fault tolerance, shouldn't you have dual fibre paths?

[Serviceguard for Linux]

Have you tried EMS (Event Monitoring Service). I'm not familiar with set-up details, but I can tell from the docs that it is supposed to be able to monitor disks and force failover.

[Dan Hoyt]

Good point Simon. This is on my dev server and we only purchased one Fibre card.

My prod box has two, but I'd still like some way to failover in case something crazy happens like both fibre cables get cut.