Monitor of Serviceguard packages with LVM 2.x-volumes

[Danny Petterson - DK]

Hi Gurus!

On our old clusters se use resls as a SERVICE to monitor the VG's (LVM 1.0), to failover if they loose connection.

Now we are implementing a new cluster, using LVM 2.1, but ouch, lo and behold, the monitor does not support LVM 2.x - which makes it impossible to use resls for in the SERVICE-script.

Any ideas on how to do this? What is common practice "these days" using the new LVM-versions?

Thanks in advance.

Kind regards
Danny Petterson

[Danny Petterson - DK]

...anyone....? Please?

;-)

[SoorajCleris]

Hi ,

In the document High Availability Monitors Version
A.04.20.11.05, A.04.20.23.05, and
A.04.20.31.05 Release Notes
HP-UX 11i v1, HP-UX 11i v2, HP-UX 11i v3

it is mentioned that

NOTE: The disk monitor does not support the following:
â ¢ HP-IB disks
â ¢ HP-FL disks
â ¢ LVM 2.0 volume groups that are available in HPâ UX 11i v3


Regards,
Sooraj

[SoorajCleris]

docs.hp.com/en/B5736-90086/B5736-90086.pdf

[Danny Petterson - DK]

Hi!

Yes, I'm painfully aware of this, that is why I'm asking what common practice would be to monitor LVM 2.x...

Kind regards
Danny

[Wim Rombauts]

You talk about a cluster , then that should be ServiceGuard, right ?

When you add the activation of a VolumeGroup to a ServiceGuard package, the availability of th VG will be a mandatory prerequisite to enable the package (and all software or functionality with it).

Maybe I don't have a clear view on your configuration and my suggestion could be useless, but why don't you just use ServiceGuard to monitor the VG ? We already have LVM 2.2 VolumeGroup monitored bij ServiceGuard 11.19. No issues.

[Danny Petterson - DK]

Hi Wim!

Thanks a bunch for your answer.

Yet, it is Serviceguard 11.19 we are using here.

In the "older" clusters we have,running LVM 1.0 and earlier versions of Serviceguard we use "resls" in a SERVICE-script for the monitoring of VG's, making the package switch to another node if the unexpected should happen, that a cluster looses all its redundant paths to the mirrored storage.

So I am very interested - how do you monitor LVM 2.2-volumegroups, making the package fail over if it looses all connections?

Again, thanks for your reply, I appreciate it a lot.

Greetings from
Danny

[Danny Petterson - DK]

Hi!

I found a workaround, making a new service_cmd using plain ls instead of resls (if ls hangs, then the disk is unaccessable).

Thans for your replys.

Greetings
Danny