[Pacemaker] Resource failover without writing to CIB

[Sam Gardner]

Hi All -

I've run through the Clusters from Scratch guide and have a decent grasp on
how to set up basic functionality such that resources failover on node
failure (ie, if node-A goes down, all resources are shunted over to node-B).

I have a high-level concept that I fear I am failing to grasp.

What I am trying to figure out now is how to failover resources if that
resource goes down (ie, on node-A I see that httpd is down, so the httpd
resource is immediately failed over to node-B). In my current configuration
(two NICS in rrp:active), httpd is simply restarted on the original node
after the monitor interval has passed.

The ocf:heartbeat:ethmonitor resource agent does this by writing the status
of the monitored nic to the CIB via attrd_updater. A constraint is then
written to monitor that CIB attribute, which then handles the failover.

Is there any way to simply monitor the response of an arbitrary ocf monitor
call, and immediately fail the affected resource over?

Thanks for any help,
Sam
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20131008/43097759/attachment-0003.html>