On Thu, Jan 29, 2009 at 20:42, Alexander Timofeev <[email protected]> wrote: > All, > > I have recently faced with strange behavior of the CRM. I have OCF compliant > RA ( ocf-tester considers it to be such ) . > It is supposed to fail over to other node on the very first failure and it > does. I noticed that my resource has fail counter set to INFINITY on node > from which it has failed over. And crm_mon report it's start as a failed > action. I have looked into logs and found that for some reason "start" was > called again after "monitor" returned OCF_ERR_GENERIC and "stop" has > successfully executed. > I supposed that after "monitor" returns error HA will call "stop" once. > After that the PE should re-calculate scores according to fail counter value > and stickinesses and so on... and decide what action should be done. > Instead "start" was called on failed resource and resource was fenced. I had > to run crm_resource -C manually to allow my resource run again on this node. > > Could anybody suggest me how I could debug this to find out what's going on?
This was fixed in http://hg.clusterlabs.org/pacemaker/stable-1.0/rev/9919f48d3313 Alas the release of 1.0.2 is taking longer than hoped due to other contractual priorities. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
