Red Hat Enterprise Linux 5 cluster node configured with fence_scsi reports SCSI reservation conflicts and during boot reports "scsi_reserve: [error] cman does not appear to be running"

Issue

• scsi_reserve reports "cman does not appear to be running" in /var/log/messages during boot, and later there are SCSI reservation conflicts:

Apr  2 11:16:37 node1 scsi_reserve: [error] cman does not appear to be running

• When mounting file systems on shared storage on a cluster node configured with fence_scsi, the mount attempt fails and there are SCSI reservation conflicts in the logs:

# service gfs2 start
Mounting GFS2 filesystems:  /sbin/mount.gfs2: mount point already used or other mount in progress
/sbin/mount.gfs2: error mounting lockproto lock_dlm
                                                           [FAILED]

Apr  2 12:41:57 node1 kernel: sd 6:0:0:23: reservation conflict
Apr  2 12:41:57 node1 multipathd: 71:16: reinstated 
Apr  2 12:41:57 node1 kernel: sd 6:0:0:23: Unhandled error code
Apr  2 12:41:57 node1 multipathd: 360050768018186abf800000000000142: remaining active paths: 1