Thanks for the quick reply. Yes, my cluster.conf contains this line (required in two node cluster configuration). <cman two_node="1" expected_votes="1"> </cman> There's also a fencing messages on the surviving node, saying that the failed node should be restarted before any recovery can proceed. Waiting for the node to rejoin the cluster or manual acknowledgement... Does this mean that I cannot use the surviving node to host my resources while my other node is fenced? Clustat says that my cluster is still qourate even with one node remaining. Please enlighten me. Thanks, Antonio Rodriguez -----Original Message----- From: David Teigland [mailto:teigland@xxxxxxxxxx] Sent: Wednesday, May 25, 2005 4:06 PM To: Rodriguez, Antonio Cc: linux-cluster@xxxxxxxxxx Subject: Re: RHEL4AS w/ cvs RHEL4 branch On Wed, May 25, 2005 at 03:43:01PM +0800, Rodriguez, Antonio wrote: > Hi everyone, > > > > I manage to install RHEL4 cvs branch on a two server RHEL4 w/ kernel > 2.6.9-5.ELsmp using the usage.txt as my guide. > > Failover seems fine when following the shutdown procedure. > > But, when simulating a broken server scenario by just turning off power > to the server, the surviving node hangs > > until such time that I was able to bootup and run "fence_tool join" > again on the previously failed node. Do you have this in your cluster.conf? <cman two_node="1" expected_votes="1"> </cman> If not, then the live node will lose quorum when you kill the other node and it won't do anything until the failed node has rejoined the cluster. If you already have that in cluster.conf, then is there a manual fencing message in /var/log/messages on the live node? Dave -- Linux-cluster@xxxxxxxxxx http://www.redhat.com/mailman/listinfo/linux-cluster
