[Pacemaker] behaviour of failure-timeout

[Andrew Beekhof]

On Fri, Sep 14, 2012 at 11:24 AM, Kashif Jawed Siddiqui
<kashifjs at huawei.com> wrote:
> Hi,
>
>>>I could not find any detailed explanation in the doc, how
>>>"failure-timeout" behaves, can someone clarify that?
> failure-timout will clear the failcount if it has increased.(i.e the resources have failed)

Strictly speaking, only 1.1 does this.
In 1.0 the cluster just ignores the failcount after it goes unchanged
for the given period of time.

> The cluster is primarily event driven but can have effect based on time. This timout can be specified as cluster property known as cluster-recheck-interval. The default is 15 mins
>
>>>My rough understanding so far is, that after a failcount is increased,
>>>pacemaker "waits for the failure-timeout" to expire and then checks if
>>>the failure condition is still on. If not, it will reset the failcount
>>>on that node. Now
> Yes
>
>>>- How does pacemaker check that, is it using a monitor operation?
> As mentioned above, depends on any events or cluster-recheck-interval.
>>>- are there re-checks at later times
> Same as above
>>>- Are checks only run on the node where the failcount was increased or
>>>on all
> No. It is run on all nodes
>
> Regards,
> Kashif Jawed Siddiqui
>
>
>>>Cheers!
>>>Mario
>
>
> _______________________________________________
> 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
> _______________________________________________
> 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