** Description changed:

- I have a cluster which is using a isolated back-end network for
- clustering-related communication (iSCSI, cman, CLVM, etc). Instead of
- using the hostnames of the nodes for the node names, I am configuring
- the nodes to all use "hostname-internal". I have an /etc/hosts file on
- each node that resolves "hostname-internal" to an IP address on the
- backend network (for all valid values of hostname). And on each host,
- /etc/default/cman sets NODENAME="hostname-internal" correctly.
+ When a cluster is brought up with enough nodes to be quorate but without
+ all nodes, it seems that the fence_tool join in /etc/init.d/cman will
+ hang until it times out.
  
- However, when I boot up my nodes, the cman init script always hangs on
- the fence_tool join -w step until it times out. However, the cluster is
- definitely up and quorate, and when the boot process finishes,
- cman_services says that fenced is running on all nodes.
- 
- I know I could just override the timeout for that particular step, but
- I'd like to actually solve the problem. Could the problem be related to
- not using the nodes' hostnames as their nodenames?
+ I stopped experiencing this problem when I removed the extra node from
+ my cluster.conf.

-- 
fence_tool times out on joining fence group
https://bugs.launchpad.net/bugs/295092
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to