Description of problem: after reboot node command mount -t gfs /dev ... enter to sleep state and stop activity on all nodes dmseg: Lock_Harness <CVS> (built Jul 15 2004 15:18:47) installed GFS <CVS> (built Jul 15 2004 15:18:29) installed CMAN <CVS> (built Jul 15 2004 15:31:01) installed NET: Registered protocol family 31 DLM <CVS> (built Jul 15 2004 15:31:12) installed Lock_DLM (built Jul 16 2004 20:15:21) installed CMAN: Waiting to join or form a Linux-cluster CMAN: sending membership request CMAN: got node c2.310.ru CMAN: got node c0.310.ru CMAN: got node c4.310.ru CMAN: got node c1.310.ru CMAN: got node c5.310.ru CMAN: got node master.310.ru CMAN: quorum regained, resuming activity dlm: gfs01: recover event 2 (first) dlm: gfs01: add nodes dlm: got connection from 1 dlm: got connection from 3 dlm: got connection from 4 dlm: got connection from 5 nodes: [root@c3 root]# cat /proc/cluster/nodes Node Votes Exp Sts Name 1 1 2 M master.310.ru 2 1 2 M c4.310.ru 3 1 2 M c1.310.ru 4 1 2 M c5.310.ru 5 1 2 M c2.310.ru 6 1 2 M c3.310.ru 7 1 2 M c0.310.ru services at rebooted node: [root@c3 root]# cat /proc/cluster/services Service Name GID LID State Code Fence Domain: "default" 0 2 join S-1,280,7 [] DLM Lock Space: "gfs01" 9 3 join S-6,20,6 [1 4 3 2 5 6] services at all other nodes: [root@c4 root]# cat /proc/cluster/services Service Name GID LID State Code Fence Domain: "default" 0 2 join S-1,80,7 [] DLM Lock Space: "gfs01" 9 3 update U-4,1,6 [1 2 3 5 4 6] GFS Mount Group: "gfs01" 10 4 run - [1 2 3 5 4] Version-Release number of selected component (if applicable): kernel 2.5.7 cluster from cvs
could you retry this one too? this looks like you may not have been doing something correctly with fencing, not necessarily a bug.
Updates with the proper version and component name.