On Fri, Feb 5, 2010 at 2:16 PM, Sanjeev Neve <[email protected]> wrote: > Hi Andrew, > > I have a constraint here and need to move to this version.
Unless you're on SLES10, whoever imposed this constraint needs their head examined. > > I have a two nodes with a group having 10 resources. The requirement > is that if any of the resources fails more than three times, we should > failover to the other node. After failing a resource I see that the > score of the resource changes (after 3 failures it becomes less than > the its score on the other node). But The group's score does not > change. It remains constant. I am not able to comprehend how group > scores are calculated. Can you please help? Not really, no. Please install a supported version (which has the benefit of being possible to use sanely). If you must use 2.1.4, some googling of the list archives will probably turn up some relevant information. > Also, is there a reference for the cib.xml syntax somewhere for version 2.0? 2.0 refers to the feature set, not the software version. > > Thanks and Regards, > Sanjeev > >> Hi, >> >> On Wed, Feb 03, 2010 at 12:36:31AM +0530, Sanjeev Neve wrote: >>> Hello All, >>> >>> I am trying to migrate a Linux-HA cluster from >>> cib_feature_revision="1.3" to cib_feature_revision="2.0". ?(heartbeat >>> 2.0.6 to heartbeat 2.1.4) The failover happens if one of the nodes is >>> stopped. However, If the resources fail, the failover does not occur. >>> The scores are allocated in such a way that if a resource fails on a >>> node more than 3 times, it should failover. >> >> If you're migrating then please use pacemaker 1.0.x with >> Heartbeat or openais/corosync. You won't get much support for >> 2.1.4. > > Score calculations was one of the things we specifically concentrated > in for pacemaker 1.0 > failure-stickiness was one of those things that sounded good in theory > but sucked in practice. > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems > _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
