Description of problem: Fenced sometimes hangs indefinitely on bootup
Version-Release number of selected component (if applicable):
How reproducible: Have not determined exactly how to reproduce. We have seen
this a number of times on different clusters. Most often we see it happen
after some change to the backend storage -- ie. changing FC HBA's, adding
LUN's, etc. When this condition occurs, the only way to recover is to bring
the node up in single user mode, then disable the cluster daemons, and go to
run level 5. At this point you can then start the cluster daemons as usual
and it seems to work fine.
Steps to Reproduce:
Is it possible to get more information about this at all?
I suspect the configuration changes are a red herring but they might be
co-incidental with something else, eg the whole cluster being rebooted after
When it next happens can you check for fencing messages in /var/log/messages?
Also, if there are other nodes in the cluster when it happens can you post the
output of "cman_tool services" from them please ?
I'll continue to try to reproduce it here.
Have we seen this one again? If not, can we close this one out?
Have not seen this in a long time. Feel free to close as invalid.
Closing as invalid, feel free to reopen the defect if it should occur again.