Topic
1 reply Latest Post - ‏2013-01-23T19:10:28Z by gpfs@us.ibm.com
mduff
mduff
34 Posts
ACCEPTED ANSWER

Pinned topic CNFS Fail Back

‏2013-01-23T01:32:55Z |
Hello all,

There has been on-going discussions for the most practical way to force a node to fail back after a CNFS fail over.

Given that you have two nodes in the CNFS cluster, and one of them has a temporary network problem and the filesystems have failed over to the other node. The temporary network issue is then fixed. What is now the recommended (and most effective) way to fail back the filesystems to the original (primary) node?

We have used mmshutdown/mmstartup on the problem node, but have not had much success. Additionally, rebooting the problem node does not always work either. We have also been told not to use mmnfsrecovernode and mmnfsnodeback.

The most effective way for us has been to reboot both nodes. Which, of course, has to be done during a maintenance window.

Thank you.
Updated on 2013-01-23T19:10:28Z at 2013-01-23T19:10:28Z by gpfs@us.ibm.com
  • gpfs@us.ibm.com
    gpfs@us.ibm.com
    96 Posts
    ACCEPTED ANSWER

    Re: CNFS Fail Back

    ‏2013-01-23T19:10:28Z  in response to mduff
    I believe that when the failed node is brought back, cNFS should re-claim its ip addrs. If this is not the case for you, please open a problem with service so it can be investigated further.