[Pacemaker] Transition-graph of pengine does a loop when configuration set order of stonith.

[renayama19661014 at ybb.ne.jp]

Hi,

We constituted two simple nodes of the clone resource of one clone resource and
stonith.

But we set order of a stonith resource and the clone resource.

We generated a problem in the next procedure.

Step1)Start two nodes and send trac1386.clone.crm.
[root at srv01 ~]# crm_mon -1 
============
Last updated: Mon Nov 29 11:29:32 2010
Stack: Heartbeat
Current DC: srv02 (faaf520d-6509-4633-8994-77d2c22c0fe3) - partition with
quorum
Version: 1.0.10-b2e39d318fda501e2fcf223c2d039b721f3679a9
2 Nodes configured, unknown expected votes
2 Resources configured.
============

Online: [ srv01 srv02 ]

 Clone Set: clnDummy
     Started: [ srv01 srv02 ]
 Clone Set: clnStonith
     Started: [ srv01 srv02 ]

Step2)Stop heartbeat in two nodes.
[root at srv01 ~]#  service heartbeat stop
Stopping High-Availability services: 

But, two nodes do not stop till a timer of the escalation starts.
There seems to be a problem in a loop doing the transition-graph that pengine
creates then.

It is necessary for us to set order of a stonith resource and the clone
resource. 

 * hb_report attached it to Bugzilla.
  * http://developerbugs.linux-foundation.org/show_bug.cgi?id=2529

Best Regards,
Hideo Yamauchi.