On Mon, May 16, 2011 at 8:27 PM, Lars Marowsky-Bree <l...@suse.de> wrote: > On 2011-05-16T09:55:13, Andrew Beekhof <and...@beekhof.net> wrote: > >> >> Couldn't both sides start shooting each other until one looses the token? >> > >> > Not more than before, though I'd expect a side that freshly rebooted to >> > come up w/o any token. >> > >> > In general, the possibility for a fencing deathmatch is not different >> > than before, though. >> I disagree. >> Before this, ignoring 2-node clusters for a moment, only one side has >> quorum which is normally a requirement for fencing to begin. > > I'm not quite sure I follow. Regular fencing is completely the same as > before - i.e., nodes will be fenced in the same situations as before. > > Only when a ticket is revoked or lost does the deadman-fencing kick in, > but that is mostly separate from quorum? > >> With this, its more like no-quorum-policy=ignore where even a single >> node can start shooting. > > A single node - or in general, a non-quorate partition - can only start > shoting _itself_ when it loses a ticket. I'm not sure I follow what > you're trying to discuss here; maybe you can rephrase?
I think it was along the lines of a partition with a token splitting into two. Would not both sides think they have the token and start shooting? _______________________________________________ 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
