On Wed, Aug 18, 2010 at 5:01 PM, Alain.Moulle <[email protected]> wrote: > Hi Dejan > > just some last questions on this subject : > > 1/ who has the responsability of the pcmk Stack ? > > is it someone from Pacemaker ? > or Oracle ?
I imagine its a joint effort. They provide an API for supplying ocfs2 with membership data and we provide the glue that implements it. > > 2/ in fact, if you re-consider my tests described in first msg, > it is not completely stalled, it succeeds to write on both sides > but one side is far far slower than the other as if the Locks > was most of the time taken by one node and the other > can take it much less often ... whereas I recall that the same > test but on OCFS2 internal stack (so without Pacemaker) is > working fine and much more load-balanced ... > > Is there some parameters to tune somewhere to get quite equivalent > load of writes on FS on both nodes ? Not in pacemaker, that would be an ocfs2 thing. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
