On Fri, Sep 30, 2011 at 8:25 AM, Mark Smith <m...@bumptechnologies.com> wrote: >> Try no-quorum-policy=freeze instead. > > I considered this option, but that puts us into a situation where if > node X and Y fail, then resources from them won't be started up on Z. > I would like to (if possible) avoid that -- I want one node to be able > to take on everything. > > I realize this may be a pipe dream.
Yep. You cant have it both ways. Either Z confirms X and Y are dead (with stonith) and starts their services, or it doesn't and it doesn't :-) > It is impossible to determine, > from the point of view of Z, whether Z has failed or whether both X > and Y have failed. This is one reason for me to turn off STONITH -- > if Z has failed, at least it won't go murder X and Y. > >> Did a stop action fail? > > Uncertain. I'll see if I can reproduce and try to record this data. > > Thanks for the response. > > > -- > Mark Smith // Operations Lead > m...@bumptechnologies.com > _______________________________________________ Pacemaker mailing list: Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
