Topic
4 replies Latest Post - ‏2012-05-07T15:04:05Z by chr78
chr78
chr78
130 Posts
ACCEPTED ANSWER

Pinned topic clarification needed on the Backward compatibility problem in GPFS 3.4.0.12

‏2012-05-05T14:56:32Z |
is the issue also present in remote cluster configurations ?
I.e. a 3.3 cluster that does a remote mount of 3.4.0.12 filesystem ?

thanks.
Updated on 2012-05-07T15:04:05Z at 2012-05-07T15:04:05Z by chr78
  • dlmcnabb
    dlmcnabb
    1012 Posts
    ACCEPTED ANSWER

    Re: clarification needed on the Backward compatibility problem in GPFS 3.4.0.12

    ‏2012-05-06T02:45:01Z  in response to chr78
    This has nothing to do with filesystem versions. This is a code compatibility problem with 3.4.0.12 nodes interacting with any node that is running 3.4.0.6 or older code. Install 3.4.0.13 instead of 3.4.0.12 while doing a rolling migration from 3.3 or old 3.4 code versions, or have clusters running different versions of code.
    • chr78
      chr78
      130 Posts
      ACCEPTED ANSWER

      Re: clarification needed on the Backward compatibility problem in GPFS 3.4.0.12

      ‏2012-05-07T07:37:38Z  in response to dlmcnabb
      Dan, I should indeed have written "remote cluster version" and not file system version.
      Just to fully understand your answer - the problem also exists in remote cluster
      configurations where e.g. a 3.3 cluster connects to a cluster containing 3.4.0.12 nodes.

      thanks again.
      • dlmcnabb
        dlmcnabb
        1012 Posts
        ACCEPTED ANSWER

        Re: clarification needed on the Backward compatibility problem in GPFS 3.4.0.12

        ‏2012-05-07T14:33:36Z  in response to chr78
        Correct, if any node is the cluster (local or remote) is running 3.3, or 3.4.0.6 or earlier, then it can cause the node running 3.4.0.12 to die with an assertion failure, and in rare cases cause filesystem corruption.
  • chr78
    chr78
    130 Posts
    ACCEPTED ANSWER

    Re: clarification needed on the Backward compatibility problem in GPFS 3.4.0.12

    ‏2012-05-07T15:04:05Z  in response to chr78
    thanks alot, Dan!