[Pacemaker] Master Slave

Andrew Beekhof andrew at beekhof.net
Mon Jul 26 06:08:36 UTC 2010


On Fri, Jul 23, 2010 at 5:30 PM, Freddie Sessler <nanogger at gmail.com> wrote:
> Dan thanks for you reply and the documentation. That all makes sense. I
> guess what I am confused about how to tell pacemaker to start mysql on both
> servers. My limited experience so far has been that pacemaker starts up
> mysql on the primary and in the event of a failure it will move the virtual
> ip to secondary node and start up the mysql service. The problem is that
> mysql is running in master/slave mode for replication and therefore needs to
> be running all the time on the secondary. Do I need a clone resource for
> this? Thanks again.

master/slave resources are a specialization of clones.

> F.
>
> On Fri, Jul 23, 2010 at 3:33 AM, Dan Frincu <dfrincu at streamwide.ro> wrote:
>>
>> First take a look at this
>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> It contains all you need for this kind of setup. I'm not aware if the M/S
>> relationship extends to other resources than DRBD, but in this case you
>> don't actually need a M/S relationship (from my point of view).
>>
>> 1. Read in the document above about 'symmetric-cluster'
>> 2. Based on results from point 1, set the resources as being allowed
>> either to run anywhere or not, and add location constraints for each
>> primitive.
>> 3. Define mandatory ordering constraints for the mysql primitive, with a
>> higher priority on the node you wish to use as a Primary node and a lower
>> priority on the Secondary node (if Primary node fails, the resource will
>> failover to Secondary node)
>> 4. Define colocation so that VIP will always run where the mysql resource
>> runs.
>> 5. (optional) define mail alerts when resource failure occurs, define what
>> the mysql resource should do when the Primary node recovers, should it
>> remain on the Secondary node and be manually moved, should it go back to the
>> Primary node by itself, these are things you might want to consider, but are
>> not mandatory to the question asked, therefore the "optional" at the
>> beginning)
>>
>> Regards,
>> Dan
>>
>> Freddie Sessler wrote:
>>
>> I have a quick question is the Master Slave setting in pacemaker only
>> allowed in regards to a DRBD device? Can you use it to create other Master
>> Slave relationships? Does all resource agents potentially involved in this
>> need to be aware of the Master Slave relationship? I am trying to set up a
>> pair fo mysql servers One is replicating from the other(handled within
>> mysql's my.cnf.) I basically want to fail over the VIP of the primary node
>> to the secondary node(which also happens to be the mysql slave) in the event
>> that the primary has its mysql server stopped. I am not using DRBD at all.
>> My config looks like the following.
>> node $id="0cd2bb09-00b6-4ce4-bdd1-629767ae0739" sipl-mysql-109
>> node $id="119fc082-7046-4b8d-a9a3-7e777b9ddf60" sipl-mysql-209
>> primitive p_clusterip ocf:heartbeat:IPaddr2 \
>> params ip="10.200.131.9" cidr_netmask="32" \
>> op monitor interval="30s"
>> primitive p_mysql ocf:heartbeat:mysql \
>> op start interval="0" timeout="120" \
>> op stop interval="0" timeout="120" \
>> op monitor interval="10" timeout="120" depth="0"
>> ms ms_mysql p_mysql \
>> meta master-max="1" master-node-max="1" clone-max="2" clone-node-max="1"
>> location l_master ms_mysql \
>> rule $id="l_master-rule" $role="Master" 100: #uname eq sipl-mysql-109
>> colocation mysql_master_on_ip inf: p_clusterip ms_mysql:Master
>> property $id="cib-bootstrap-options" \
>> stonith-enabled="false" \
>> no-quorum-policy="ignore" \
>> start-failure-is-fatal="false" \
>> expected-quorum-votes="2" \
>> symmetric-cluster="false" \
>> dc-version="1.0.9-89bd754939df5150de7cd76835f98fe90851b677" \
>> cluster-infrastructure="Heartbeat"
>> rsc_defaults $id="rsc-options" \
>> resource-stickiness="100"
>>
>> What's happening is that mysql is never brought up due to the following
>> errors:
>> ul 22 16:15:07 sipl-mysql-109 pengine: [22890]: info: native_color:
>> Resource p_mysql:0 cannot run anywhere
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: info: native_color:
>> Resource p_mysql:1 cannot run anywhere
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: info:
>> native_merge_weights: ms_mysql: Rolling back scores from p_clusterip
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: info: master_color:
>> ms_mysql: Promoted 0 instances of a possible 1 to master
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: info: native_color:
>> Resource p_clusterip cannot run anywhere
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: info: master_color:
>> ms_mysql: Promoted 0 instances of a possible 1 to master
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: notice: LogActions: Leave
>> resource p_clusterip (Stopped)
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: notice: LogActions: Leave
>> resource p_mysql:0 (Stopped)
>> Jul 22 16:15:07 sipl-mysql-109 pengine: [22890]: notice: LogActions: Leave
>> resource p_mysql:1 (Stopped)
>>
>> I thought I may have overcome this with my location and colocation
>> directive but it failed. Could someone give me some feedback on what I am
>> trying to do, my config and the resulting errors?
>> Thanks
>> F.
>>
>> ________________________________
>> _______________________________________________
>> 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://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>>
>>
>> --
>> Dan FRINCU
>> Systems Engineer
>> CCNA, RHCE
>> Streamwide Romania
>>
>> _______________________________________________
>> 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://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>>
>
>
> _______________________________________________
> 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://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>
>




More information about the Pacemaker mailing list