[Pacemaker] A processor joined or left the membership and a new membership was formed.

[Andrew Beekhof]

On Sun, Jul 1, 2012 at 7:09 PM, Hugo Deprez <hugo.deprez at gmail.com> wrote:
> Thank you for the information I'll change the configuration and see
> what will happen.
> Hard to troubleshoot network issue.
> I didn't detect any packets losts.
>
> Running cluster on VM seems a bit tricky.

What might be happening is that the OS is pausing the VM(s), albeit
very briefly.
That might be enough to make it look like there were network hiccups.

http://bugzilla.redhat.com/show_bug.cgi?id=820821 has some related discussion.

>
> On 28 June 2012 04:42, Andrew Beekhof <andrew at beekhof.net> wrote:
>> On Sun, Jun 24, 2012 at 10:55 PM, Hugo Deprez <hugo.deprez at gmail.com> wrote:
>>> Hello,
>>>
>>> I guess this as laready been raised but I do have the following
>>> message on some of my clusters :
>>>
>>> Jun 24 12:00:02 server corosync[27089]:   [TOTEM ] A processor joined
>>> or left the membership and a new membership was formed.
>>> Jun 24 14:10:02 server corosync[27089]:   [TOTEM ] A processor joined
>>> or left the membership and a new membership was formed.
>>>
>>> Don't really know why. Ressources Load and ram seems to be fine on that server.
>>
>> More than likely its a network issue of some kind.
>>
>>> I guess I need to take care of this message as this can lead to wrong
>>> behaviour on the cluster.
>>>
>>> I was wondering If I need to change the value of  consensus (currently
>>> 3000) to solve this issue.
>>
>> More likely either or both of these two would help more:
>>
>>         # How long before declaring a token lost (ms)
>>         token:          5000
>>
>>         # How many token retransmits before forming a new configuration
>>         token_retransmits_before_loss_const: 10
>>
>> _______________________________________________
>> 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