On Fri, 10 Feb 2006 11:14:57 -0500 Greg Forte <gforte@xxxxxxxxxxxxxxxxxxx> wrote: > If you have a two-node cluster, and you don't want both services running > on the same node, then if one node goes down your whole cluster is > defunct. So ... what's the problem with using "run exclusively"? I > realize this locks each service to a particular node, but if you don't > want them running on the same node then there's no functional difference > between that and what you want for a two-node cluster. Yes, I fully understand that, but my situation is a bit more complicated. I have to do some tests to see if I'm going to go with "run exclusively" or I'm going to hack init scripts. Since I've dumped GFS for performance reasons, I went with one node having locally mounted volume and the other node mounting it from the first one via NFS. It's obvious that I want these two mounts be always on different nodes. Now the applications that are running atop of these are another story. Some, like postgres, are going to be tied to the "local" mount, others, like apache, can live anywhere. But I first have to sort out the local vs. nfs mount issue ... -- Jure Pečar http://jure.pecar.org -- Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
