[Pacemaker] ERROR with startup-fencing

[Andrew Beekhof]

On Nov 27, 2008, at 3:33 AM, Satomi TANIGUCHI wrote:

> Hi,
>
> I got the following ERROR when I started up only one node in the  
> cluster
> with startup-fencing="true".

You don't need to set that option, true is the default value.

>
>
> crmd[8984]: 2008/11/20_13:57:15 ERROR: te_graph_trigger: Transition  
> failed:
> terminated
>
> And the other node is not STONITH'ed when this ERROR occurs.

Based on the case 1 graph, I'd say the PE has a bug.
Can you create a bug for this please?

>
>
> I tried 4 cases.
> And this ERROR occurs only when
> the cluster has grouped and cloned STONITH plugins
> and <nodes> in cib.xml has all nodes' information.
> See the table in attached txt.
> And I attached hb_report's archive file in the ERROR case, and graph  
> files
> of each case.
>
> [The environment]
> * Heartbeat    : 9dbbc2cbd414
> * Pacemaker1.0 : 2ed0b0ebaf57
> * The cluster has 2 nodes.
>
> [Test-process]
> (1) Delete all files in /var/lib/heartbeat/crm
> (2) Put the cib.xml in the directory above.
> (3) Started up only one node in the cluster
>
> Just for reference,
> I also tried Case1. with the following process, then the same ERROR  
> appeared.
> (1') Delete all files in /var/lib/heartbeat/crm
> (2') Started up only one node in the cluster
> (3') cibadmin -U -x cib.xml
>
>
> Is this a bug?
>
>
> Best Regards,
> Satomi TANIGUCHI
> [Tested Cases]
>
> +---------+--------------------------------------- 
> +-------------------------+-----------------------+
> | CaseNo. | 	 STONITH setting in cib.xml 	  | <nodes> has nodes'  
> info |  <nodes> has no info	|
> +=========+======================================= 
> +=========================+=======================+
> |	   1. | <resources> 						  | ERROR!! 				| no error. 			|
> |		  |   <clone>							  | the other node is not	| the other node is  
> not |
> |		  | 	<group> 						  | STONITH'ed. 			| STONITH'ed.			|
> |		  | 	  <primitive class="stonith" ...> | 						|						|
> +---------+--------------------------------------- 
> +-------------------------+						|
> |	   2. | <resources> 						  | no error.				|						|
> |		  |   <clone>							  | the other node is		|						|
> |		  | 	<primitive class="stonith" ...>   | STONITH'ed. 			|						|
> +---------+---------------------------------------+ 						|						|
> |	   3. | <resources> 						  | 						|						|
> |		  |   <group>							  | 						|						|
> |		  | 	<primitive class="stonith" ...>   | 						|						|
> +---------+---------------------------------------+ 						|						|
> |	   4. | <resources> 						  | 						|						|
> |		  |   <primitive class="stonith" ...>	  | 						|						|
> +---------+--------------------------------------- 
> +-------------------------+-----------------------+
>
> * "<nodes> has nodes' info" means:
>  </crm_config>
>  <nodes>
> 	<node uname="rh5node1" type="normal" id="286f4fcb-519e-4a23- 
> b39f-9ab0017d0442"/>
> 	<node uname="rh5node2" type="normal"  
> id="fb62f5f4-015c-466a-8778-7b5c1c5639d6"/>
>  </nodes>
>  <resources>
>
> ** "<nodes> has no info" means:
>  </crm_config>
>  <nodes/>
>  <resources>
> <hb_report.tar.bz2><pe-warn-0.CGP_case1.gif><pe- 
> warn-0.CP_case2.gif><pe-warn-0.GP_case3.gif><pe- 
> warn-0.P_case4.gif>_______________________________________________
> Pacemaker mailing list
> Pacemaker at clusterlabs.org
> http://list.clusterlabs.org/mailman/listinfo/pacemaker