Q: VA7100 (HP22): not answering requests any more? - HP UX

This is a discussion on Q: VA7100 (HP22): not answering requests any more? - HP UX ; Hello, I would think this scenario is impossible, but reality seems to know better: An VA7100 with 15 disks that was completely OK one day, stopped to respond to host requests the next day until a controller was pulled out ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: Q: VA7100 (HP22): not answering requests any more?

  1. Q: VA7100 (HP22): not answering requests any more?

    Hello,

    I would think this scenario is impossible, but reality seems to know better:
    An VA7100 with 15 disks that was completely OK one day, stopped to respond to
    host requests the next day until a controller was pulled out and re-inserted.

    FAIK the reason was a disk that was considered bad during a routine scrub, and
    thus triggered a rebuild of the array. During the rebuild there was another
    error and everything seemed to halt. I don't know exactly what was going on,
    but that was unexpected.

    Not even armdiscover was able to find the array. Maybe things did not come to
    a halt, but every request took 10 minutes or more (which doesn't make much
    difference in practice).

    What I suspect is that HP se the limits for a bad disk so conservative that
    it's much too late when the array tries to fix things (It seems in the EVA
    arrays there is a similar problem: Disks all seemed to fail immediately and
    suddenly without warning while being up). In my experience disks really
    indicate that they are having a problem before complete failure, but built-in
    S.M.A.R.T. test may be crap: I had one ATA disk that was unreadable and
    unwritable in several sectors while the SMART status was OK. Vendors just
    don't want to replace defective parts...

    After re-inserting the controller in the VA7100, the disk that was considered
    defective before seems OK now.

    Any deep insights on this?

    Regards,
    Ulrich

  2. Re: Q: VA7100 (HP22): not answering requests any more?

    Hi,

    another question on the VA7100: Is it expected that host read and write
    requests time out when the VA7100 does a rebuild at "low" priority due to a
    missing disk? This happened about one hour after a defective disk was halfway
    pulled out. At that time a "disc scrub" was also in progress (which is in
    progress for over three days now).

    Regards,
    Ulrich

+ Reply to Thread