[Pacemaker] Question about fallback of resource group

[Dejan Muhamedagic]

Hi,

On Tue, Mar 02, 2010 at 05:59:28PM +0100, philipp.achmueller at arz.at wrote:
> hi,
> 
> configuration and behavoir:
> 
> $ crm configure show
> node lnx0012a \
>         attributes standby="off"
> node lnx0012b \
>         attributes standby="on"
> primitive pingd ocf:heartbeat:pingd \
>         params host_list="10.1.236.100" multiplier="100" \
>         op monitor interval="15s" timeout="20s"
> primitive testIP ocf:heartbeat:IPaddr2 \
>         params ip="10.1.236.62" nic="eth1" \
>         op monitor interval="10s"
> group testRG pingd testIP
> property $id="cib-bootstrap-options" \
>         dc-version="1.0.5-05c8b63cbca7ce95182bb41881b3c5677f20bd5c" \
>         expected-quorum-votes="2" \
>         cluster-infrastructure="openais" \
>         stonith-enabled="false" \
>         no-quorum-policy="ignore" \
>         last-lrm-refresh="1267537804"
> 
> $ crm_mon --one-shot -V
> ============
> Last updated: Tue Mar  2 17:41:05 2010
> Stack: openais
> Current DC: lnx0012a - partition with quorum
> Version: 1.0.5-05c8b63cbca7ce95182bb41881b3c5677f20bd5c
> 2 Nodes configured, 2 expected votes
> 1 Resources configured.
> ============
> 
> Online: [ lnx0012b lnx0012a ]
> 
> Resource Group: testRG
>     pingd       (ocf::heartbeat:pingd): Started lnx0012a
>     testIP      (ocf::heartbeat:IPaddr2):       Started lnx0012a
> 
> ----------------------------
> 
> changing state of active node "lnx0012a" to standby the RG moves to 
> seconde node "lnx0012b"
> $ crm node standby lnx0012a
> 
> ---------------------------
> $ crm_mon --one-shot -V
> .....
> Node lnx0012a: standby
> Online: [ lnx0012b ]
> 
> Resource Group: testRG
>     pingd       (ocf::heartbeat:pingd): Started lnx0012b
>     testIP      (ocf::heartbeat:IPaddr2):       Started lnx0012b
> 
> ----------------------------
> 
> after reintegrate lnx0012a to the cluster the "testRG" moves back to 
> lnx0012a
> $ crm node online lnx0012a
> 
> ---------------------------
> $ crm_mon --one-shot -V
> .....
> Online: [ lnx0012b lnx0012a ]
> 
> Resource Group: testRG
>     pingd       (ocf::heartbeat:pingd): Started lnx0012a
>     testIP      (ocf::heartbeat:IPaddr2):       Started lnx0012a
> ......
> ######################################################
> 
> how to prevent falling back the RG after reintegrate first node in the 
> cluster??
> 
> ######################################################
> 
> after manual move i see the rsc_location in configuration:
> 
>     <constraints>
>       <rsc_location id="cli-prefer-testRG" rsc="testRG">
>         <rule id="cli-prefer-rule-testRG" score="INFINITY" 
> boolean-op="and">
>           <expression id="cli-prefer-expr-testRG" attribute="#uname" 
> operation="eq" value="lnx0012a" type="string"/>
>         </rule>
>       </rsc_location>
>     </constraints>
> 
> $ crm configure delete cli-prefer-testRG
> 
> but this didn't work, the RG still fall back to lnx0012a

Check documentation at clusterlabs.org and look for
resource-stickiness.

Thanks,

Dejan

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