Red Hat Knowledgebase

Manual fencing is not recommended in a production environment. Manual fencing can cause file system corruption. In addition, manual fencing requires manual intervention any time a node is fenced, reducing the availability of the cluster.

Nevertheless, sometimes manual fencing is used. In the event that a node in the Global File System (GFS) cluster is fenced, it will not be able to rejoin the quorum automatically. The following steps must be done to bring the fenced node back on-line:

1. Log into the server that fenced the node in question. If there is only one lockserver, it will be that system. If there are multiple lock servers, searching /var/log/messages on each of them should reveal which one performed the fencing.
2. Shutdown the fenced node.
3. From the lock server that did the fencing, execute fence_ack_manual -s IPAddress where "IPAddress" is the IP address of the node that was fenced. For example:

   fence_ack_manual -s 10.0.0.15

4. Reboot the node that was fenced. It should join the quorum successfully when it boots.

For more information on manual fencing, see the GFS 6.0 documentation at: http://www.redhat.com/docs/manuals/csgfs/admin-guide/s1-fence-methods.html