2.1.3 has a well known bug that was fixed in later releases. fail
count doesn't get increased when resource fails. You have to upgrade.

On Wed, Feb 18, 2009 at 11:30 AM, Pavel Georgiev <[email protected]> wrote:
> I`m using heartbeat 2.1.3, default Centos 5 rpm. I`m running 3 nodes with a
> single resource (LSB RA) which has an equal score on each server. I`m having
> two "issues" (which actually might be feature)
> 1) If I stop the resource between two heartbeat "monitor" intervals, it
> detects it is down and restarts it (which is OK) and the failure count
> becomes one. If I stop th resource it is again restarted on the same node
> and the failure count is still 1 - I cant get it to increase. Am I missing
> some configuration trick?
>
> 2) If the "start" operation of the RA fails, the score is set to -INFINITY.
> Is it possible to control so that the score is just decreased (by
> resource_failure_stickiness) so that the node is still eligible for running
> the resource?
>
> 3) I can`t clear the failure count for a resource with ` crm_failcount -D -U
> server1 -r controller_vm`; Also, how do I change the score of a node if it
> gets a "_INFINITY" score (as decribed in th eprevious issue).
>
>
> My config:
>
> <cib epoch="0" admin_epoch="0" num_updates="0">
>   <configuration>
>      <crm_config/>
>      <nodes/>
>      <resources>
>         <primitive id="controller_vm_resource" class="lsb"
> type="controller_vm" provider="applogic">
>            <operations>
>               <op id="controller_vm_resource_status" interval="60s"
> name="monitor" timeout="5s" start_delay="10s" on_fail="restart"/>
>               <op id="controller_vm_resource_start" name="start"
> timeout="10s" on_fail="restart"/>
>               <op id="controller_vm_resource_stop" name="stop"
> timeout="10s"/>
>            </operations>
>            <meta_attributes id="controller_vm_resource_attr">
>               <attributes>
>                  <nvpair id="controller_vm_resource_attr_1"
> name="resource_stickiness" value="100"/>
>                  <nvpair id="controller_vm_resource_attr_2"
> name="resource_failure_stickiness" value="-100"/>
>               </attributes>
>            </meta_attributes>
>         </primitive>
>      </resources>
>      <constraints>
>         <rsc_location id="run_controller_vm_resource"
> rsc="controller_vm_resource">
>            <rule id="pref_run_controller_service_resource_1" score="1000">
>               <expression id="rule_controller_vm_1" attribute="#uname"
> operation="eq" value="server1"/>
>            </rule>
>            <rule id="pref_run_controller_service_resource_2" score="1000">
>               <expression id="rule_controller_vm_2" attribute="#uname"
> operation="eq" value="server2"/>
>            </rule>
>            <rule id="pref_run_controller_service_resource_3" score="1000">
>               <expression id="rule_controller_vm_3" attribute="#uname"
> operation="eq" value="server3"/>
>            </rule>
>         </rsc_location>
>      </constraints>
>   </configuration>
> </cib>
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>



-- 
Serge Dubrouski.
_______________________________________________
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