Help with failing drive in striped XLV - SGI

This is a discussion on Help with failing drive in striped XLV - SGI ; I'm not real familiar with IRIX, so forgive my ignorance. We have some old archives that I am either trying to get off the disk (preferably) or a NetWorker tape restore. I have a IRIX 6.2 system with several striped ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: Help with failing drive in striped XLV

  1. Help with failing drive in striped XLV

    I'm not real familiar with IRIX, so forgive my ignorance. We have some
    old archives that I am either trying to get off the disk (preferably)
    or a NetWorker tape restore. I have a IRIX 6.2 system with several
    striped XLV volumes, and of the 8 volumes, 5 are not mounting. The
    volumes are striped across 3 different SCSI controllers, with one drive
    from each controller making up the stripe set. On one controller,
    drives 2-6 all seem to have issues. For starters, when the drive shelf
    is powered on, and each drive spins up (one at a time) with the light
    on, each of these drives' light stays on, indicating to me there is
    some sort of problem, while the 3 drives that are part of other XLV
    volumes operate fine and the volumes are mounted.

    When running xlv_mgr, the 5 volumes with issues all show up with a
    "Label entry xlv2.data.0.0.0 has newer timestamp" message, and each
    shows up [clean], and it sees all 3 drives, while the rest are
    [active]. So I thought since it seems to be able to see the drives in
    some form, and I can see the vol header in dvhtool, etc., maybe
    deleting the xlv object and recreating it in the exact sequence would
    get me going again. Problem is, when I go to do this in xlv_mgr, I get
    "xlv[13628]: Unable to write label on /dev/rdsk/dks3d3vh Failed in
    xlv_lower with I/O error". Finally, when I do an
    fx->exercise->sequential, I get I/O errors on block 0.

    Please tell me...am I hopelessly screwed, or are there methods for
    determining if my data is still intact and retrievable by me or if I
    should start looking at some tape restores. If I've skipped over steps
    or am ahead of myself, please shed some light for me.

    Thanks,
    Dave


  2. Re: Help with failing drive in striped XLV

    In article <1162513376.715311.316810@h54g2000cwb.googlegroups. com>,
    Kingsman wrote:
    >get me going again. Problem is, when I go to do this in xlv_mgr, I get
    >"xlv[13628]: Unable to write label on /dev/rdsk/dks3d3vh Failed in
    >xlv_lower with I/O error". Finally, when I do an
    >fx->exercise->sequential, I get I/O errors on block 0.


    >Please tell me...am I hopelessly screwed, or are there methods for
    >determining if my data is still intact and retrievable by me or if I
    >should start looking at some tape restores.


    You will need a tape restore for any volume that relied on the
    presence of /dev/rdsk/dks3d3s* .

    What RAID level did you use? If you just striped without
    any redundancy, then you might have increased performance but you cubed
    the probability of error.

  3. Re: Help with failing drive in striped XLV


    Walter Roberson wrote:
    > In article <1162513376.715311.316810@h54g2000cwb.googlegroups. com>,
    > Kingsman wrote:
    > >get me going again. Problem is, when I go to do this in xlv_mgr, I get
    > >"xlv[13628]: Unable to write label on /dev/rdsk/dks3d3vh Failed in
    > >xlv_lower with I/O error". Finally, when I do an
    > >fx->exercise->sequential, I get I/O errors on block 0.

    >
    > >Please tell me...am I hopelessly screwed, or are there methods for
    > >determining if my data is still intact and retrievable by me or if I
    > >should start looking at some tape restores.

    >
    > You will need a tape restore for any volume that relied on the
    > presence of /dev/rdsk/dks3d3s* .
    >
    > What RAID level did you use? If you just striped without
    > any redundancy, then you might have increased performance but you cubed
    > the probability of error.


    Yes, I understand RAID levels, and I would not have setup a striped
    RAID if it were up to me, but this is an inherited system with archival
    information only, so that's what I'm stuck with. Appreciate the
    help...was hoping someone might have some lower-level disk tools that I
    might attempt, since it seems the drives have some sort of
    functionality.


  4. Re: Help with failing drive in striped XLV

    In article <1162583208.948301.67430@h48g2000cwc.googlegroups.c om>,
    Kingsman wrote:

    >Yes, I understand RAID levels, and I would not have setup a striped
    >RAID if it were up to me, but this is an inherited system with archival
    >information only, so that's what I'm stuck with.


    If it was a physical RAID of level higher than 1 or if you had
    XLV plexes for logical RAIDs, then replacing the failing dks3s3
    would give a possibility of rebuilding the RAID. Though from what
    you said, it might not be the drive itself that is failing.

    Lemme think... XLV volumes are self-identifying in the volume header,
    so if you move 3s3 to a different controller, XLV should pay attention.
    If fx the drive on a different controller still gives I/O errors
    then the drive itself is a probably the problem.


+ Reply to Thread