Description of problem: I had one machine in a cluster come up with the gulm init script on and a valid cluster.conf file, and it ended up hanging because I never started gulm on any of the other nodes in the cluster. Starting ccsd: [ OK ] Starting cman:[WARNING] [WARNING]lustered mirror log:[WARNING] Starting lock_gulmd: [HANG] It appears that if a master is never found then it will hang. We need to have it timeout eventually continue booting the node. wait_for_master() { i=0 rtrn=0 stoptime=$(($SECONDS + $GULM_QUORUM_TIMEOUT)) while [ $GULM_QUORUM_TIMEOUT -eq 0 -o $SECONDS -lt $stoptime ] do find_master rtrn=$? case $rtrn in 0) break ;; # master was found 1) ;; # master was not found 2) break ;; # gulm_tool error 3) break ;; # we are expired esac sleep 5 i=$(($i+1)) done return $rtrn } Version-Release number of selected component (if applicable): [root@link-08 ~]# rpm -q gulm gulm-1.0.7-0
I've been unable to replicate this with Corey, moving to NEEDINFO until we see this again.
Have not been able to reproduce this, will reopen if seen again.