On Wed, May 20, 2009 at 6:18 PM, Mikael Kermorgant <
[email protected]> wrote:

> Hello,
>
> I have a drbd + service group (filesystem + zope + ip) configured with the
> latest pacemaker on two nodes (zeo1 and zeo2).
>
> It works quite well, failover works as expected when putting active node
> standby or by shutting the nic (thanks to pingd), but I'd like to have only
> one automatic failover.
>
> Zeo1 should be the master. If something happens to zeo1, failover to zeo2
> occurs. But if zeo1 returns to normal conditions, it should not be restored
> automatically as master.
>
> There's a rule giving zeo1 a score of 100.
>       <rsc_location id="ms-drbd0-master-on-zeo1" rsc="ms-drbd0">
>         <rule id="ms-drbd0-master-on-zeo1-rule" role="master" score="100">
>           <expression attribute="#uname"
> id="ms-drbd0-master-on-zeo1-expression" operation="eq" value="zeo1"/>
>         </rule>
>       </rsc_location>
>
>
> I thought a resource-stickiness with a value higher than 100 put on either
> resource ms-drbd0 or the service group would hinder the return ro zeo1, but
> that's not the case.
>
> But my tests show me it doesn(t work that way.
>

Reply to myself :

I put a resource-stickiness of INFINITY on my group and restarted heartbeat
services on both nodes, reload the conf from scratch and it works.

Maybe it was a too low score, maybe something else...

Regards,

Mikael
_______________________________________________
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