[Pacemaker] detecting resource failures after maintenance

[Andrew Beekhof]

On 11/05/2013, at 1:53 AM, Jeffrey Lewis <jlewis at 42lines.net> wrote:

> It seems pacemaker is not properly detecting resource failures after
> maintenance.  Example follows.
> 
> Pacemaker is managing two IPaddr2 resources.  Both resources are
> online, and all is well.
> 
> jlewis at qa3db22:~$ sudo crm resource show
> shard0_ip (ocf::heartbeat:IPaddr2) Started
> shard1_ip (ocf::heartbeat:IPaddr2) Started
> 
> I decide to do some maintenance and set is-managed-default=false.
> This way, pacemaker will continue monitoring all resources, but will
> not take action should a resource fail.
> 
> jlewis at qa3db22:~$ sudo crm configure property is-managed-default=false
> 
> jlewis at qa3db23:~$ sudo crm resource show
> shard0_ip (ocf::heartbeat:IPaddr2) Started  (unmanaged)
> shard1_ip (ocf::heartbeat:IPaddr2) Started  (unmanaged)
> 
> I then take resource 'shard1_ip' offline using ifconfig.  Pacemaker
> correctly notices that this resource has failed.
> 
> jlewis at qa3db23:~$ sudo ifconfig eth0:shard1 down
> 
> jlewis at qa3db23:~$ sudo crm resource show
> shard0_ip (ocf::heartbeat:IPaddr2) Started  (unmanaged)
> shard1_ip (ocf::heartbeat:IPaddr2) Started  (unmanaged) FAILED
> 
> However, when I set is-managed-default=true, pacemaker incorrectly
> think resource 'shard1_ip' is ok, but the IP address is still down.

Version of pacemaker? Logs? Config? (ie. failure timeout)

> 
> jlewis at qa3db23:~$ sudo crm configure property is-managed-default=true
> 
> jlewis at qa3db23:~$ sudo crm resource show
> shard0_ip (ocf::heartbeat:IPaddr2) Started
> shard1_ip (ocf::heartbeat:IPaddr2) Started
> 
> I don't necessarily expect pacemaker to start this IP, since it was
> stopped when pacemaker was not managing this resource, but I do expect
> pacemaker to correctly report current status.
> 
> Any hints?
> 
> Thanks,
> Jeffrey
> 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org