Device goes busy, yet missing - Aix

This is a discussion on Device goes busy, yet missing - Aix ; Hello, Folks, First, let me apologize for not having complete information. We're stymied and we're turning to anyone who might have a useful insight into what condition may exist when we see this. Does anyone here know under what circumstances ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: Device goes busy, yet missing

  1. Device goes busy, yet missing

    Hello, Folks,

    First, let me apologize for not having complete information. We're
    stymied and we're turning to anyone who might have a useful insight into
    what condition may exist when we see this.

    Does anyone here know under what circumstances one might see this:

    # lsvg -o
    0516-304 : Unable to find device id 00c2c87000004c000000011afed768b6 in
    the Device
    Configuration Database.
    rootvg

    #lsvg
    rootvg

    ....and we only see this from lsvg after this output is received from a
    vendor-supplied script (no, I don't know what the script is trying to
    do):

    Method error (/etc/methods/chgpowerdisk):
    0514-062 Cannot perform the requested function because the
    specified device is busy.
    0516-1320 recreatevg: Unable to recreate volume group.

    This is looney-tunes! How can a busy device not exist???!!!

    Here's the kicker: a reboot clears the condition!

    Again, apologies for the scant information.

    Any insights anyone has would indeed be most welcome! (Flames will not
    be helpful; so, flame-suppression is requested.)

    D.J.D.

  2. Re: Device goes busy, yet missing

    On Jul 10, 3:44 am, David J Dachtera
    wrote:
    > Hello, Folks,
    >
    > First, let me apologize for not having complete information. We're
    > stymied and we're turning to anyone who might have a useful insight into
    > what condition may exist when we see this.
    >
    > Does anyone here know under what circumstances one might see this:
    >
    > # lsvg -o
    > 0516-304 : Unable to find device id 00c2c87000004c000000011afed768b6 in
    > the Device
    > Configuration Database.
    > rootvg
    >
    > #lsvg
    > rootvg
    >
    > ...and we only see this from lsvg after this output is received from a
    > vendor-supplied script (no, I don't know what the script is trying to
    > do):
    >
    > Method error (/etc/methods/chgpowerdisk):
    > 0514-062 Cannot perform the requested function because the
    > specified device is busy.
    > 0516-1320 recreatevg: Unable to recreate volume group.
    >
    > This is looney-tunes! How can a busy device not exist???!!!


    Could it be that you installed powerpath on a system which uses
    already the EMC?
    If yes read the: PowerPath for AIX Installation and Administration
    Guide .

    My bet is that you installed powerpath without varyoff the volume
    groups whom hdisk devices are going to be converted to powerpath
    devices.

    hth
    Hajo


  3. Re: Device goes busy, yet missing

    On 07/10/2008 03:44 AM, David J Dachtera wrote:
    > # lsvg -o
    > 0516-304 : Unable to find device id 00c2c87000004c000000011afed768b6 in
    > the Device Configuration Database.
    > rootvg


    We recently had the same error (also with PowerPath). It was caused by
    duplication of PVIDs - check the output of lspv.

    See:
    IY88164: DUPLICATION OF PVID IN VGDA BY RUNNING MKVG OR EXTENDVG APPLIES
    TO AIX 5200-10
    IY88240: DUPLICATION OF PVID IN VGDA BY RUNNING MKVG OR EXTENDVG APPLIES
    TO AIX 5300-06

    The way I fixed it requires you know the proper hdisk name for
    "00c2c87000004c000000011afed768b6", i.e. from previous lspv output.
    (Warning: if you mess this up, it could of course cause management to
    enter your office with heavy submachine guns)

    But let's assume you know from older lspv output that the PVID was
    associated with the EMC path that the O/S knows as 'hdisk29', and that
    'hdisk29' currently has the wrong PVID.

    Backup your current ODM:
    odmget CuAt > backup.odm

    Simply create a file containing:
    CuAt:
    name = "hdisk29"
    attribute = "pvid"
    value = "00c2c87000004c000000011afed768b6"
    type = "R"
    generic = "D"
    rep = "s"
    nls_index = 2

    And run:
    odmchange -o CuAt -q "name=hdisk29 AND attribute=pvid" ./odm.hdisk29.new

    An alternative solution might be to apply the APAR, reboot and hope for
    the best...

    cheers
    Niel

  4. Re: Device goes busy, yet missing

    Hajo Ehlers wrote:
    >
    > On Jul 10, 3:44 am, David J Dachtera
    > wrote:
    > > Hello, Folks,
    > >
    > > First, let me apologize for not having complete information. We're
    > > stymied and we're turning to anyone who might have a useful insight into
    > > what condition may exist when we see this.
    > >
    > > Does anyone here know under what circumstances one might see this:
    > >
    > > # lsvg -o
    > > 0516-304 : Unable to find device id 00c2c87000004c000000011afed768b6 in
    > > the Device
    > > Configuration Database.
    > > rootvg
    > >
    > > #lsvg
    > > rootvg
    > >
    > > ...and we only see this from lsvg after this output is received from a
    > > vendor-supplied script (no, I don't know what the script is trying to
    > > do):
    > >
    > > Method error (/etc/methods/chgpowerdisk):
    > > 0514-062 Cannot perform the requested function because the
    > > specified device is busy.
    > > 0516-1320 recreatevg: Unable to recreate volume group.
    > >
    > > This is looney-tunes! How can a busy device not exist???!!!

    >
    > Could it be that you installed powerpath on a system which uses
    > already the EMC?
    > If yes read the: PowerPath for AIX Installation and Administration
    > Guide .
    >
    > My bet is that you installed powerpath without varyoff the volume
    > groups whom hdisk devices are going to be converted to powerpath
    > devices.


    Well, that's close, in a nuke-bomb sort of way...

    Here's what I have to help flesh this out:

    The backup strategy calls for BCVs (clones, really) to be activated
    while the database (Oracle) is quiesced (in backup mode). The clones are
    then imported to another LPAR and the database backup is run off using
    RMAN+NetWorker.

    The issue seems to arise post the backup step where possibly, as you
    suggest, a varyoff is being missed, or is perhaps colliding with another
    operation possibly due to a timing issue.

    There are over 120 clones to process (Raw LVs). So, it's done in some
    kind of loop. Speculation is that possibly a short sleep is needed
    between loop iterations to prevent collisions between SYMCLIAPI
    operations.

    We're still digging. So, thanx for the clue. We recruited the vendors to
    script this in the hope of expedience. Hope we didn't shoot our own
    feet... (We were previously a VMS shop where this runs very smoothly,
    and my UN*X skills go back to AT&T System-III circa. 1986, but not much
    since then.)

    D.J.D.

+ Reply to Thread