[Pacemaker] Resource switching again and again ...

[Andrew Beekhof]

On 05/04/2013, at 3:24 AM, Jerome BRUSQ <jerome.brusq at signalis.com> wrote:

> Hi all,
> I’m starting using Pacemaker and Corosync, and I find it very nice, it’s a very good tool  !
>  
> I have create a cluster (2 nodes),  and pacemaker monitors a group of  3 resources :
> - a custom service (lsb script), that  starts 2 processes ( 1 application + 1 archiving )
> - a gateway (ocf:pacemaker:ping)
> - a virtual interface (ocf:heartbeat:IPAddres)
>  
> For my custom service I set :
>  Migration-threshold=3
> Failure-timeout=500
>  
> That is working well, after 3 failures, the group is moving to the other node.
>  
> My problem:
>  
> Context : The resources are running on node1. My process (launched by my custom lsb script)  is archiving on a NAS.
>  
> My custom service status is KO  (archiving process is KO because the NAS is full for example), pacemaker tries to restart 3 times my service .. and finally move the group on node2. -> normal behavior
>  
> Unfortunately, node2 uses the same NAS. So, pacemaker tries to restart 3 times my service and move back to node 1…
> Etc etc …
>  
> Questions :
> -          Is there a way to prevent switching back again and again  ( in my example, even if archiving processes are KO, application process is OK, so I would like application process still being up).
> The behavior I am looking for is  : after moving resources to node 2 , do not authorize any migration for the next  hour (even if node 2 has a failure)  …

Try setting Failure-timeout=0 for the custom service.
Any failures will need to be cleared out manually and the service will no longer be able to run after exhausting all available nodes.

>  
> Thanks for your help
>  
> jerome
>  
>  
>  
>  
>  
> _______________________________________________
> 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