Hi Thanks but ... OK I think my english is not clear enough ... I tried to explain why Pacemaker was fencing (because I forced it to fence by avoiding heartbeat to transit) and why this configuration Pacemaker and ocfs2.pcmk was not roobust at all (again: it was at the end of 2010) ... but you misunderstood my message I think ... sorry for my english ... Alain
De : "Ulrich Windl" <[email protected]> A : "General Linux-HA mailing list" <[email protected]> Date : 02/02/2012 16:43 Objet : Re: [Linux-HA] Antw: Status about ocfs2.pcmk ? Envoyé par : [email protected] >>> <[email protected]> schrieb am 02.02.2012 um 16:27 in Nachricht <of98c1ec8f.f2ff97ef-onc1257998.00530432-c1257998.0053a...@bull.net>: > Hi > Don't remember in details, it was at the end of 2010 ... but : > Why pacemaker is fencing a node ? AFAIK mostly because of timeouts, which, in turn, may result from high delays (maybe I/O delays). > because it was one of my simple HA test : for example, make the heartbeat > no more working so > that pacemaker fences the node, and it did it for sure, but on the > remaining node, sometimes ocfs2 crashes > the node, or the FS ocfs2 was passing in read-only ... On a two-node-cluster you typically have a STONITH deathmatch. Probably it's working as designed. Regards, Ulrich > I did simple HA tests, such as this one above, to check the robustness of > the configuration and many times, > it failed and make the HA cluster down. > That's why I ask for status today. > > Alain > > > > De : "Ulrich Windl" <[email protected]> > A : "General Linux-HA mailing list" <[email protected]> > Date : 02/02/2012 16:01 > Objet : Re: [Linux-HA] Antw: Status about ocfs2.pcmk ? > Envoyé par : [email protected] > > > >>>> <[email protected]> schrieb am 02.02.2012 um 16:05 in Nachricht > <of5716529b.0ca482e6-onc1257998.0050e240-c1257998.00518...@bull.net>: > > Hi > > Thanks. > > OK but I also could mount the FS , the problems in 2010 on RHEL was that > > > the configuration was not robust , meaning that during validation tests, > > > there were often cases where both nodes were dead : one fenced by > > Pacemaker and the other killed itself by ocfs2, or problems on mount > > read-only FS after failover , etc. > > Hi! > > I'd start inspecting the logs: Why is pacemaker fencing a node? Why is the > filesytem read-only? > > Regards, > Ulrich > > > > Alain > > > > > > > > De : "Ulrich Windl" <[email protected]> > > A : <[email protected]> > > Date : 02/02/2012 15:33 > > Objet : [Linux-HA] Antw: Status about ocfs2.pcmk ? > > Envoyé par : [email protected] > > > > > > > > Hi! > > I have something running using OCFS on SLES11 SP1: > > > > ocf:pacemaker:controld > > ocf:ocfs2:o2cb > > > > At least I could mount the filesystem with it: > > /dev/drbd_r0 on /exports/ocfs/samba type ocfs2 > > (rw,_netdev,acl,cluster_stack=pcmk) > > > > Regards, > > Ulrich > > > >>>> <[email protected]> schrieb am 02.02.2012 um 14:54 in Nachricht > > <offed8db74.9970c9e4-onc1257998.004a64db-c1257998.004b1...@bull.net>: > > > Hi > > > > > > Just wonder if someone has succeded to configured a working HA > > > configuration with Pacemaker/corosync > > > and OCFS2 file systems, meaning using ocfs2.pcmk , on RHEL6 mainly > (and > > > eventually SLES11) ? > > > > > > (I tried at the end of 2010 but gave up after a few weeks because it > was > > > > > not working at all) > > > > > > Thanks if someone can give a status? > > > Regards > > > Alain Moullé > > > _______________________________________________ > > > 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 > > _______________________________________________ > > 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 > _______________________________________________ > 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 _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
