I think now is a good time to mention the quorum enhancements going into 3.4. I believe we'll have a way to determine quorum and resolve split-brain even in replica 2 setups. This will be possible via the creation of an arbiter node, which isn't a separate server, and whose sole job will be to determine quorum and determine the direction for self-heal. I'll leave it to the devs to comment on how much is implemented now and whether it's in the master branch. -JM ----- Original Message ----- > Hi! > > > -----Urspr?ngliche Nachricht----- > > Von: Brian Candler [mailto:B.Candler at pobox.com] > > Gesendet: Freitag, 16. November 2012 09:28 > > An: Martin Emrich > > Cc: gluster-users at gluster.org > > Betreff: Re: Avoid Split-brain and other stuff > > > > > That would be perfectly acceptable, as long as it would heal > > > deterministically (last one wins, or renamed conflicting files) > > > > Not for me it wouldn't. "Last one wins" means "one set of updates > > thrown > > away", i.e. definite data loss, which will be compounded when > > further > > updates take place. > > > > Automatic renaming means either that the file vanishes from its > > original name > > (so the application which looks for the file breaks anyway), or > > that one version > > has the original name and the other version is renamed - which can > > also result > > in irrepairable damage. > > For my use case (File server), that would be highly unlikely, and > acceptable in rare cases. > Of course it would be a "No-Go" for computing applications or > database storage. > > > What you don't want is both nodes to be up, both reachable only by > > a subset > > of clients, and updates occurring on both. > > Hmm, that's exactly what I want (Branch offices temporarily > disconnected can continue to work), thus GlusterFS is probably not > the right tool for me. > > Ciao > > Martin > > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://supercolony.gluster.org/mailman/listinfo/gluster-users >
