Hi Or perhaps you can also choose to set the meta parameter failure-timeout to xx secondes, so that resource could migrate back after a failover in case of failure again ...
failure-timeout (Default : 0=disabled) : How many seconds to wait before acting as if the failure has not occured, and potentially allowing the resource back to the node on which it failed Alain Le 18/04/2013 09:34, fabian.herschel a écrit : >>>> Actually now, I'm looking for configuration/parameters that will stop the >>>> running node >>> Shut it down, stop the cluster software on it, or just stop any other >>> resources there? >>> >> If any resource fails in node1, stop the all resources in node 1 and start >> all the resources on node2 > For all resources, which need to be migrated together you either could define > them in a group (implicit colocated, but also ordered), or you define > explicit colocation constraints with score +INFINITY. > > For all resources which should migrate on the first failure you could set the > fail-count limit to 1. > > But again at least in a 2-node cluster such a construct has the disadvantage, > that after the services has failed on all nodes it will not started any more. > > What I do not know by hart, but you can check that, if "migrate" is also a > valid on-fail action for the monitor operation. If that´s the case forget the > failcount-part above and only implement the monitor-operation with the > on-fail-action "migrate". > > Hope this helps > Fabian > > > > Von Samsung-Tablet gesendet > _______________________________________________ > Linux-HA mailing list > Linux-HA@lists.linux-ha.org > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems