[Linux-HA] heartbeat 2.0.2 + stonith

Dave Blaschke debltc at us.ibm.com
Tue Oct 4 10:43:48 MDT 2005


announces at genos.es wrote:

>El mar, 04-10-2005 a las 17:38, Dave Blaschke escribió:
>  
>
>>announces at genos.es wrote:
>>
>>    
>>
>>>Thanks. Now I have modified the cib.xml which uses an example stonith
>>>plugin:
>>>...................
>>>
>>>        <primitive id="lampara_stonith" class="stonith"
>>>type="external/test">
>>>          <instance_attributes>
>>>            <attributes>
>>>              <nvpair name="ST_HOSTLIST" value="heartbeat1" />
>>>            </attributes>
>>>          </instance_attributes>
>>>        </primitive>
>>>
>>>.....................
>>>
>>>Now heartbeat runs without errors, but, when I simulate an error
>>>condition (a resource which doesn't stop), it doesn't fence the node. I
>>>can read the following in log files:
>>>
>>>
>>>pengine[3326]: 2005/10/04_16:36:58 info: mask(stages.c:stage6):
>>>Scheduling Node heartbeat1 for shutdown
>>>pengine[3326]: 2005/10/04_16:36:58 WARN: mask(stages.c:stage6):
>>>Scheduling Node heartbeat1 for STONITH
>>>
>>>
>>>But It doesn't call my external/test stonith plugin. 
>>>
>>>Later I can read:
>>>
>>>stonithd[15626]: 2005/10/04_16:09:49 info: Failed to STONITH the node
>>>heartbeat1: optype=1, op_result=2
>>> 
>>>
>>>      
>>>
>>This message indicates that the ST_GENERIC_RESET operation failed with 
>>S_ACCESS, usually indicating that the STONITH device cannot be 
>>accessed.  Are you sure that there's an executable file test in the 
>>/usr/lib/stonith/plugins/external/ directory that can process all of the 
>>commands that Alan listed below?
>>
>>    
>>
>
>Yes, the script runs and it has in the first line:
>
>echo $@ >> /tmp/heartbeat.log
>
>And I see how stonithd calls the plugin. When resources run in the first
>node, I have in the log file:
>
>
>getconfignames
>status
>gethosts
>  
>
How about if you run:

stonith -d -t external/test -p "heartbeat1" -T reset heartbeat1

from the command line and capture the output.  This should provide some 
interesting debug output...

>
>_______________________________________________
>Linux-HA mailing list
>Linux-HA at lists.linux-ha.org
>http://lists.linux-ha.org/mailman/listinfo/linux-ha
>See also: http://linux-ha.org/ReportingProblems
>
>
>  
>




More information about the Linux-HA mailing list