Thank you. We intend to upgrade to pacemaker, but unfortunately we have
to continue to support version 2 right now.
Regarding the dampen parameter, I have tried tuning it, but as long as
all cluster nodes are using the same dampening value there is still a
race condition where the first node to report the score change gets the
resource moved to it.
If the node currently running the resource had a smaller dampening value
then maybe the resource would stay where it is.

Regards,
Brad

Andrew Beekhof wrote:
> For starters you could try not using 2.0.8
>
> On Fri, Jan 14, 2011 at 9:48 PM, Brad Johnson <[email protected]> wrote:
>   
>> Running heartbeat 2.0.8 I have a problem with unnecessary moving of our
>> resource to a different node when a pingd node goes dead or comes back
>> up. I am talking about the ping node itself going dead, so ALL cluster
>> nodes lose connectivity to it.
>> There seems to be a race condition depending on which node detects and
>> reports the ping node status change first.
>>     
>
> Thats what the dampen parameter is for.  You might need to tune it.
> Or there might be a bug that got fixed some time in the last four years.
>
>   
>> If the node which is not
>> running the resource reports the status change first, then temporarily
>> the ping score is unbalanced and the resource is moved. It seems to me
>> there should be some way of configuring some sort of cluster delay
>> between getting a ping node status change and moving resources to
>> another node.
>>
>> Thanks,
>> Brad
>>
>> _______________________________________________
>> Linux-HA mailing list
>> [email protected]
>> http://lists.linux-ha.org/mailman/listinfo/linux-ha
>> See also: http://linux-ha.org/ReportingProblems
>>
>>     
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>   
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to