[Pacemaker] strange behavior with stonith-action

[zorg]

Hello
I have a working example using pacemaker corosync with stonith

primitive s_ipmi_hv1 stonith:fence_ipmilan \
     op monitor interval="20s" \
     params action="off" pcmk_host_check="static-list" 
pcmk_host_list="hv2" ipaddr="21.37.43.132" login="admin" passwd="xxxxx" 
power_wait="4" lanplus="true" delay="1" verbose="1"
primitive s_ipmi_hv2 stonith:fence_ipmilan \
     op monitor interval="20s" \
     params action="off" pcmk_host_check="static-list" 
pcmk_host_list="hv1" ipaddr="21.37.43.130" login="admin" passwd="xxxx" 
power_wait="4" lanplus="true" delay="10" verbose="1"
location loc-prefer-stonith-1 s_ipmi_hv1 \
     rule $id="loc-prefer-stonith-1-rule" inf: #uname eq hv1
location loc-prefer-stonith-2 s_ipmi_hv2 \
     rule $id="loc-prefer-stonith-2-rule" inf: #uname eq hv2


But When I set
stonith-action=poweroff stonith stop working

and if I reset back to
stonith-action=reboot it working killing the node

I really can't undestand why

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clusterlabs.org/pipermail/pacemaker/attachments/20130121/7ac4b048/attachment-0002.html>