To follow up on this matter, I've solved the problem and it was not a bug in pacemaker.
fencing/commands.c checks to see if it can fence the node with fence_legacy. If this fails commands.c will disable checking for the remainder of the pacemaker session and throw an error to syslog when the stonith op is requested. My problem was that perl was broken and fence_legacy was failing. Since commands.c only checks whether fencing works once it made tracking down fence_legacy as the problem very difficult. I now have working stonith using pacemaker 1.1.6. Many thanks for your debugging suggestions. -Hal On Wed, Nov 30, 2011 at 2:30 PM, Hal Martin <[email protected]> wrote: > I do not know where this timeout is coming from. It appears to be > independent of any cluster settings. > > I've tested causing a node to be marked UNCLEAN by messing with a > VirtualDomain RA. > > Expected results: STONITH > Observed results: Timeout. No STONITH. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
