FW: Recovery at a DR Site

  • 1.  FW: Recovery at a DR Site

    Posted 02-25-2005 09:26 AM
    Hi All,
    We ran a second DR test this week, and once again I somehow got a data
    snapshot during the time a journal was offloading at the main data center.
    This caused a journal to show as ""condensing"" once the regions warmstarted
    at the DR site. The odds of twice getting a data snapshot during the few
    minutes a journal is offloading is amazing, although I think this may be
    because this client may have predefined ""VARY JOURNAL"" times and the data
    snapshots were done at the same time both weeks.
    Anyway, this time when I got to the DR site, one journal was in the
    condensing phase. I ran a manual journal offload with the RESTART parameter
    and the journal finished condensing successfully.
    For those who are intimate with the journal offload/condense process, my
    question becomes the following: if you do DR by having your data ""mirrored""
    at a DR site, is there a way I can get caught with unrecoverable data in the
    journals if the disaster occurred while a journal was offloading? Or is the
    journal offload process so smart it can handle anything?

    Joe Lupico
    IDMS System Support
    ""Our World is a Happy World""

    IDMS Public Discussion Forum


    Re: Recovery at a DR Site
    You're under a slight misunderstanding. We are not mirroring the tape
    activity, just the DASD activity. On this particular client, the journal
    offload tapes are never taken offsite. They used to use a process of
    transmitting the journal offload data to a remote site and then performing a
    roll forward, but that has just been replaced by a process where all the
    DASD is kept in (very close) sync.
    Basically, the offloaded journal data really no longer serves a useful
    function, except for auditing.
    If a journal offload is occuring at the main data center, I am never going
    to see the tape at the DR site. All I've got are the disk journals.
    However, my concern is the state of the disk journal which was in the
    condense phase at the main data center when it was ""snapshot"" to the DR
    site. In last week's test, I ran the offload at DR with a RESTART and
    everything seemed fine.
    I can rephrase the question this way. If you lose your entire mainframe
    while a journal offload is performing a condense, when you restart the IDMS
    system, is there a possibility that should the system need to read the old
    journal (the one which was being condensed) in order to roll out any
    in-flight run units, the WARMSTART could fail?

    Joe Lupico
    IDMS System Support
    ""Our World is a Happy World""