3 Replies Latest reply on May 27, 2012 11:54 AM by 911168

    Disaster recovery of remaining OVM server


      I hope some of you can help me with some disaster recovery tests I have to do shortly.

      The environment has 2 OVM servers: servre1 and server2 in a server pool.
      Server1 has the cluster filesystem shared to it via multiipathing, and in turn, server1 shares the repository and cluster fs with server2 via NFS.

      In the meantime, there are other LUNS shared with just server2, which are being replicated with the repository and cluster fs that is currently being used.

      If server1 "dies" and this is easy to simulate as just shutting it down, it will stop sharing the repository and fs with server2, the server pool will have errors, and so will server 1 (it cannot be contacted), and so will server2 (it has lost access to repository and cluster fs).

      Oracle documentation explains very clearly how to add and remove server pools and servers neatly, but it makes no mention of how to clean up a server in order to reuse in on a new server pool, with a new repository, in this case, the replicated LUNs that were only in server2.

      Has any of you done something similar to this? Do I have to go into the OVM Manager database at row level to delete every mention of the old repository? Is there a way of cleaning up the server2 for reuse (other than reinstalling)? In OVM 2.2 there was a cleanup.py script, but I didn´t find such script in 3.0.3...

      Any suggestions on how to test this will be welcome. If I have overlooked some useful documentation on this, please point me to it...