[Pacemaker] Question about the error when fencing failed

Andrew Beekhof andrew at beekhof.net
Wed Apr 17 05:03:56 UTC 2013


This should solve your issue:

	https://github.com/beekhof/pacemaker/commit/dbbb6a6

On 11/04/2013, at 7:23 PM, Kazunori INOUE <inouekazu at intellilink.co.jp> wrote:

> Hi Andrew,
> 
> (13.04.08 11:04), Andrew Beekhof wrote:
>> 
>> On 05/04/2013, at 3:21 PM, Kazunori INOUE <inouekazu at intellilink.co.jp> wrote:
>> 
>>> Hi,
>>> 
>>> When fencing failed (*1) on the following conditions, an error occurs
>>> in stonith_perform_callback().
>>> 
>>> - using fencing-topology. (*2)
>>> - fence DC node. ($ crm node fence dev2)
>>> 
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:   notice: handle_request: Client crmd.2282.b9e69280 wants to fence (reboot) 'dev2' with device '(any)'
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:   notice: handle_request: Forwarding complex self fencing request to peer dev1
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:     info: stonith_command: Processed st_fence from crmd.2282: Operation now in progress (-115)
>>> Apr  3 17:04:47 dev2 pengine[2281]:  warning: process_pe_message: Calculated Transition 2: /var/lib/pacemaker/pengine/pe-warn-0.bz2
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:     info: stonith_command: Processed st_query from dev1: OK (0)
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:     info: stonith_action_create: Initiating action list for agent fence_legacy (target=(null))
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:     info: stonith_command: Processed st_timeout_update from dev1: OK (0)
>>> Apr  3 17:04:47 dev2 stonith-ng[2278]:     info: dynamic_list_search_cb: Refreshing port list for f-dev1
>>> Apr  3 17:04:48 dev2 stonith-ng[2278]:   notice: remote_op_done: Operation reboot of dev2 by dev1 for crmd.2282 at dev1.4494ed41: Generic Pacemaker error
>>> Apr  3 17:04:48 dev2 stonith-ng[2278]:     info: stonith_command: Processed st_notify reply from dev1: OK (0)
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: crm_abort: stonith_perform_callback: Triggered assert at st_client.c:1894 : call_id > 0
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result   <st-reply st_origin="stonith_construct_reply" t="stonith-ng" st_rc="-201" st_op="st_query" st_callid="0" st_clientid="b9e69280-e557-478e-aa94-fd7ca6a533b1" st_clientname="crmd.2282" st_remote_op="4494ed41-2306-4707-8406-fa066b7f3ef0" st_callopt="0" st_delegate="dev1">
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result     <st_calldata>
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result       <st-reply t="st_notify" subt="broadcast" st_op="reboot" count="1" src="dev1" state="4" st_target="dev2">
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result         <st_calldata>
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result           <st_notify_fence state="4" st_rc="-201" st_target="dev2" st_device_action="reboot" st_delegate="dev1" st_remote_op="4494ed41-2306-4707-8406-fa066b7f3ef0" st_origin="dev1" st_clientid="b9e69280-e557-478e-aa94-fd7ca6a533b1" st_clientname="crmd.2282"/>
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result         </st_calldata>
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result       </st-reply>
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result     </st_calldata>
>>> Apr  3 17:04:48 dev2 crmd[2282]:    error: stonith_perform_callback: Bad result   </st-reply>
>>> Apr  3 17:04:48 dev2 crmd[2282]:  warning: stonith_perform_callback: STONITH command failed: Generic Pacemaker error
>>> Apr  3 17:04:48 dev2 crmd[2282]:   notice: tengine_stonith_notify: Peer dev2 was not terminated (st_notify_fence) by dev1 for dev1: Generic Pacemaker error (ref=4494ed41-2306-4707-8406-fa066b7f3ef0) by client crmd.2282
>>> Apr  3 17:07:11 dev2 crmd[2282]:    error: stonith_async_timeout_handler: Async call 2 timed out after 144000ms
>>> 
>>> Is this the designed behavior?
>> 
>> Definitely not :-(
>> Is this the first fencing operation that has been initiated by the cluster?
> 
> Yes.
> I attached crm_report.
> 
>> Or has the cluster been running for some time?
>> 
> 
> ----
> Best Regards,
> Kazunori INOUE
> 
>>> 
>>> *1: I added "exit 1" to reset() of stonith-plugin in order to make
>>>    fencing fail.
>>> 
>>>  $ diff -u libvirt.ORG libvirt
>>>  --- libvirt.ORG 2012-12-17 09:56:37.000000000 +0900
>>>  +++ libvirt     2013-04-03 16:33:08.118157947 +0900
>>>  @@ -240,6 +240,7 @@
>>>       ;;
>>> 
>>>       reset)
>>>  +    exit 1
>>>       libvirt_check_config
>>>       libvirt_set_domain_id $2
>>> 
>>> *2:
>>>  node $id="3232261523" dev2
>>>  node $id="3232261525" dev1
>>>  primitive f-dev1 stonith:external/libvirt \
>>>      params pcmk_reboot_retries="1" hostlist="dev1" \
>>>      hypervisor_uri="qemu+ssh://bl460g1n5/system"
>>>  primitive f-dev2 stonith:external/libvirt \
>>>      params pcmk_reboot_retries="1" hostlist="dev2" \
>>>      hypervisor_uri="qemu+ssh://bl460g1n6/system"
>>>  location rsc_location-f-dev1 f-dev1 \
>>>      rule $id="rsc_location-f-dev1-rule" -inf: #uname eq dev1
>>>  location rsc_location-f-dev2 f-dev2 \
>>>      rule $id="rsc_location-f-dev2-rule" -inf: #uname eq dev2
>>>  fencing_topology \
>>>      dev1: f-dev1 \
>>>      dev2: f-dev2
>>>  property $id="cib-bootstrap-options" \
>>>      dc-version="1.1.10-1.el6-132019b" \
>>>      cluster-infrastructure="corosync" \
>>>      no-quorum-policy="ignore" \
>>>      stonith-timeout="70s"
>>> 
>>> Best Regards,
>>> Kazunori INOUE
>>> 
>>> _______________________________________________
>>> 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://bugs.clusterlabs.org
>> 
>> 
>> _______________________________________________
>> 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://bugs.clusterlabs.org
> <unexplained-crmd-error.tar.bz2>_______________________________________________
> 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://bugs.clusterlabs.org





More information about the Pacemaker mailing list