[Pacemaker] Problems settings up MySQL failover

[Dejan Muhamedagic]

Hi,

On Fri, Feb 19, 2010 at 02:19:15PM -0300, Lior Marantenboim wrote:
> Hello everyone
> 
> I've having trouble trying to set up a simple MySQL failover (make Pacemaker
> start it on one node, and if it shuts down, start it on the other). CRM
> shows as if MySQL is not running but it actually is, and since it doesn't
> see it that way, it tries to start it on the other node as well.
> I'm using Debian Lenny, corosync 1.2.0, Pacemaker 1-0-051b78d5f321 and
> cluster glue
> 
> Thanks for any help!
> 
> This is my setup:
> 
> hpblade02:~# crm configure show
> node hpblade01
> node hpblade02 \
>     attributes standby="off"
> primitive MySQLIP ocf:heartbeat:IPaddr \
>     params ip="10.10.99.57" cidr_netmask="32" \
>     op monitor interval="30s" \
>     meta target-role="Started"
> primitive mysql ocf:heartbeat:mysql \
>     op monitor interval="30s" timeout="60s" \
>     meta target-role="Started"
> location cli-prefer-mysql mysql \
>     rule $id="cli-prefer-rule-mysql" inf: #uname eq hpblade01
> property $id="cib-bootstrap-options" \
>     dc-version="1.0.7-051b78d5f32195bd9d83447343606211993169e7" \
>     cluster-infrastructure="openais" \
>     expected-quorum-votes="2" \
>     stonith-enabled="false" \
>     no-quorum-policy="ignore" \
>     last-lrm-refresh="1266599381"
> rsc_defaults $id="rsc-options" \
>     resource-stickiness="100"
> hpblade02:~#
> 
> And here is crm_mon output:
> 
> hpblade02:~# crm_mon
> Defaulting to one-shot mode
> You need to have curses available at compile time to enable console mode
> ============
> Last updated: Fri Feb 19 14:16:30 2010
> Stack: openais
> Current DC: hpblade01 - partition with quorum
> Version: 1.0.7-051b78d5f32195bd9d83447343606211993169e7
> 2 Nodes configured, 2 expected votes
> 2 Resources configured.
> ============
> 
> Online: [ hpblade01 hpblade02 ]
> 
>  MySQLIP    (ocf::heartbeat:IPaddr):    Started hpblade01
> 
> Failed actions:
>     mysql_start_0 (node=hpblade01, call=22, rc=-2, status=Timed Out):
> unknown exec error
>     mysql_start_0 (node=hpblade02, call=24, rc=-2, status=Timed Out):
> unknown exec error

This contradicts the statement above, i.e. the start action timed
out. It may be that mysql actually started, but then monitor
failed. Don't have the details. If it takes a long time for the
database to start, then increase the start operation timeout
(defaults to 20 sec, from the default-action-timeout property).
Check the logs for any mysql messages (incl. those relayed by
lrmd). If you can't figure out what's going on, make a hb_report
and post it.

Thanks,

Dejan

> hpblade02:~#

> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker