Could disk shadowing stress SCSI? - VMS

This is a discussion on Could disk shadowing stress SCSI? - VMS ; I noticed I was getting soft errors when adding a member to a shadow set. The errors are for the disks, but they show a SCSI STATUS of CHECK CONDITION in DIAGNOSE. Could it be that the SCSI is stressed ...

+ Reply to Thread
Results 1 to 8 of 8

Thread: Could disk shadowing stress SCSI?

  1. Could disk shadowing stress SCSI?

    I noticed I was getting soft errors when adding a member to a shadow
    set.

    The errors are for the disks, but they show a SCSI STATUS of CHECK
    CONDITION in DIAGNOSE.

    Could it be that the SCSI is stressed during the period of heavy
    activity when a shadow set is being reconstituted?

    One of the disks had just ran clean for ANAL/MEDIA/EXER
    so I don't see how the errors could be from the disk.


  2. Re: Could disk shadowing stress SCSI?

    In article <1188414878.371078.292290@r34g2000hsd.googlegroups. com>,
    tadamsmar wrote:
    >I noticed I was getting soft errors when adding a member to a shadow
    >set.


    The same happens to me - the disk involved is a different physical
    configuration (1" vs. "full-height") than its partner. No other errors occur
    after the copy or merge is complete. In my many years' experience working with
    shadowed disks, this is not a cause for concern.

    >The errors are for the disks, but they show a SCSI STATUS of CHECK
    >CONDITION in DIAGNOSE.
    >
    >Could it be that the SCSI is stressed during the period of heavy
    >activity when a shadow set is being reconstituted?
    >
    >One of the disks had just ran clean for ANAL/MEDIA/EXER
    >so I don't see how the errors could be from the disk.


    My experience tells me that disks are in trouble when they throw "many" errors
    in a "short" time (for some values of "many" and "short"). Of course, since
    you are using shadowing, it will cost no more than the price of another drive
    to determine if there is indeed a problem with your "problem" disk. :-)

  3. Re: Could disk shadowing stress SCSI?

    On Aug 29, 9:23 pm, bradhamil...@comcast.net (Brad Hamilton) wrote:
    > In article <1188414878.371078.292...@r34g2000hsd.googlegroups. com>,
    >
    > tadamsmar wrote:
    > >I noticed I was getting soft errors when adding a member to a shadow
    > >set.

    >
    > The same happens to me - the disk involved is a different physical
    > configuration (1" vs. "full-height") than its partner. No other errors occur
    > after the copy or merge is complete. In my many years' experience working with
    > shadowed disks, this is not a cause for concern.
    >
    > >The errors are for the disks, but they show a SCSI STATUS of CHECK
    > >CONDITION in DIAGNOSE.

    >
    > >Could it be that the SCSI is stressed during the period of heavy
    > >activity when a shadow set is being reconstituted?

    >
    > >One of the disks had just ran clean for ANAL/MEDIA/EXER
    > >so I don't see how the errors could be from the disk.

    >
    > My experience tells me that disks are in trouble when they throw "many" errors
    > in a "short" time (for some values of "many" and "short"). Of course, since
    > you are using shadowing, it will cost no more than the price of another drive
    > to determine if there is indeed a problem with your "problem" disk. :-)


    I had another system go nuts and start repeatedly rebuilding DSA0 and
    bogging the system due to soft errors on a disk during a merge. The
    problem quit if I stopped the merge by dismounting the disk being
    merged in. Granted, only happened once and I have 50 system years of
    experience. But I am a bit concerned because I did have this problem.


  4. RE: Could disk shadowing stress SCSI?

    > -----Original Message-----
    > From: tadamsmar [mailto:tadamsmar@yahoo.com]
    > Sent: August 30, 2007 8:17 AM
    > To: Info-VAX@Mvb.Saic.Com
    > Subject: Re: Could disk shadowing stress SCSI?
    >
    > On Aug 29, 9:23 pm, bradhamil...@comcast.net (Brad Hamilton) wrote:
    > > In article <1188414878.371078.292...@r34g2000hsd.googlegroups. com>,
    > >
    > > tadamsmar wrote:
    > > >I noticed I was getting soft errors when adding a member to a shadow
    > > >set.

    > >
    > > The same happens to me - the disk involved is a different physical
    > > configuration (1" vs. "full-height") than its partner. No other

    > errors occur
    > > after the copy or merge is complete. In my many years' experience

    > working with
    > > shadowed disks, this is not a cause for concern.
    > >
    > > >The errors are for the disks, but they show a SCSI STATUS of CHECK
    > > >CONDITION in DIAGNOSE.

    > >
    > > >Could it be that the SCSI is stressed during the period of heavy
    > > >activity when a shadow set is being reconstituted?

    > >
    > > >One of the disks had just ran clean for ANAL/MEDIA/EXER
    > > >so I don't see how the errors could be from the disk.

    > >
    > > My experience tells me that disks are in trouble when they throw

    > "many" errors
    > > in a "short" time (for some values of "many" and "short"). Of

    > course, since
    > > you are using shadowing, it will cost no more than the price of

    > another drive
    > > to determine if there is indeed a problem with your "problem" disk.

    > :-)
    >
    > I had another system go nuts and start repeatedly rebuilding DSA0 and
    > bogging the system due to soft errors on a disk during a merge. The
    > problem quit if I stopped the merge by dismounting the disk being
    > merged in. Granted, only happened once and I have 50 system years of
    > experience. But I am a bit concerned because I did have this problem.


    The issue is likely related to something like SCSI cable length or impropertermination
    (flaky connections?) that results in errors on the drive furthest down the bus. As
    A previous reply stated, simply replacing the drive will tell you if you have a drive
    or SCSI bus / cable issue.

    I also seem to remember that some self test issues on older servers e.g. pwr up, left
    devices in a sort Of undetermined state that caused some issues in some configs (YMMV).

    Fwiw, I always like to Issue an additional >>> INIT before booting older systems.

    I would also ensure that you have the latest patch kits applied for the version of
    OpenVMS you are using.

    Regards


    Kerry Main
    Senior Consultant
    HP Services Canada
    Voice: 613-592-4660
    Fax: 613-591-4477
    kerryDOTmainAThpDOTcom
    (remove the DOT's and AT)

    OpenVMS - the secure, multi-site OS that just works.



  5. Re: Could disk shadowing stress SCSI?

    Brad Hamilton wrote:
    > In article <1188414878.371078.292290@r34g2000hsd.googlegroups. com>,
    > tadamsmar wrote:
    >> I noticed I was getting soft errors when adding a member to a shadow
    >> set.

    >
    > The same happens to me - the disk involved is a different physical
    > configuration (1" vs. "full-height") than its partner. No other errors occur
    > after the copy or merge is complete. In my many years' experience working with
    > shadowed disks, this is not a cause for concern.
    >
    >> The errors are for the disks, but they show a SCSI STATUS of CHECK
    >> CONDITION in DIAGNOSE.
    >>
    >> Could it be that the SCSI is stressed during the period of heavy
    >> activity when a shadow set is being reconstituted?
    >>
    >> One of the disks had just ran clean for ANAL/MEDIA/EXER
    >> so I don't see how the errors could be from the disk.

    >
    > My experience tells me that disks are in trouble when they throw "many" errors
    > in a "short" time (for some values of "many" and "short"). Of course, since
    > you are using shadowing, it will cost no more than the price of another drive
    > to determine if there is indeed a problem with your "problem" disk. :-)


    DKdriver will not do command queueing when doing data compares or some other
    error recovery techniques, and if doing read after write will also not run
    tagged queueing. This will lower disk throughput. If you are getting SCSI
    check condition a lot, dkdriver or mkdriver will not be performing as well as
    usual.

    Of course, it is also correct that when errors occur a lot, your data is well on
    the way to being toast much of the time. If your problems are SCSI bus termination
    or length or the like, of course, you may just have perpetual flakiness, but if
    that precipitates enough sector replacements for the drive to run out of spares
    you are again potentially up the estuary.

    An old wise saying: "there are PERFECTLY GOOD TECHNICAL REASONS why you must
    sacrifice a small goat to your SCSI bus every so often."

    Glenn Everhart

  6. Re: Could disk shadowing stress SCSI?

    Glenn Everhart wrote:
    > Brad Hamilton wrote:
    >
    >> In article <1188414878.371078.292290@r34g2000hsd.googlegroups. com>,
    >> tadamsmar wrote:
    >>
    >>> I noticed I was getting soft errors when adding a member to a shadow
    >>> set.

    >>
    >>
    >> The same happens to me - the disk involved is a different physical
    >> configuration (1" vs. "full-height") than its partner. No other
    >> errors occur
    >> after the copy or merge is complete. In my many years' experience
    >> working with
    >> shadowed disks, this is not a cause for concern.
    >>
    >>> The errors are for the disks, but they show a SCSI STATUS of CHECK
    >>> CONDITION in DIAGNOSE.
    >>>
    >>> Could it be that the SCSI is stressed during the period of heavy
    >>> activity when a shadow set is being reconstituted?
    >>>
    >>> One of the disks had just ran clean for ANAL/MEDIA/EXER
    >>> so I don't see how the errors could be from the disk.

    >>
    >>
    >> My experience tells me that disks are in trouble when they throw
    >> "many" errors
    >> in a "short" time (for some values of "many" and "short"). Of course,
    >> since
    >> you are using shadowing, it will cost no more than the price of
    >> another drive
    >> to determine if there is indeed a problem with your "problem" disk.
    >> :-)

    >
    >
    > DKdriver will not do command queueing when doing data compares or some
    > other
    > error recovery techniques, and if doing read after write will also not run
    > tagged queueing. This will lower disk throughput. If you are getting SCSI
    > check condition a lot, dkdriver or mkdriver will not be performing as
    > well as
    > usual.
    >
    > Of course, it is also correct that when errors occur a lot, your data is
    > well on
    > the way to being toast much of the time. If your problems are SCSI bus
    > termination
    > or length or the like, of course, you may just have perpetual flakiness,
    > but if
    > that precipitates enough sector replacements for the drive to run out of
    > spares
    > you are again potentially up the estuary.
    >
    > An old wise saying: "there are PERFECTLY GOOD TECHNICAL REASONS why you
    > must
    > sacrifice a small goat to your SCSI bus every so often."
    >
    > Glenn Everhart


    If a small goat is not immediately available, one may substitute two
    medium chickens! ;-)



  7. Re: Could disk shadowing stress SCSI?

    "Richard B. Gilbert" wrote:
    > [snip]
    > If a small goat is not immediately available, one may substitute two
    > medium chickens! ;-)


    SSSSSSSSSSSSSSHHHHHHHHHHHHHHHH !!!!!

    You weren't supposed to tell about me and that goat...

    ....not much into cluckers, either...

    --
    David J Dachtera
    dba DJE Systems
    http://www.djesys.com/

    Unofficial OpenVMS Marketing Home Page
    http://www.djesys.com/vms/market/

    Unofficial Affordable OpenVMS Home Page:
    http://www.djesys.com/vms/soho/

    Unofficial OpenVMS-IA32 Home Page:
    http://www.djesys.com/vms/ia32/

    Unofficial OpenVMS Hobbyist Support Page:
    http://www.djesys.com/vms/support/

  8. Re: Could disk shadowing stress SCSI?

    In article <46DAFA30.3040103@gce.com>,
    Glenn Everhart wrote:

    > An old wise saying: "there are PERFECTLY GOOD TECHNICAL REASONS why you must
    > sacrifice a small goat to your SCSI bus every so often."


    Right on cue, some evidence of that concept here:

    <http://www.telegraph.co.uk/news/main...09/05/wgoat105
    ..xml>

    "Goats sacrificed to help fix Nepalese plane"

    Apparently for something the size of a Boeing 757, 2 goats are required.:.)

    --
    Paul Sture

    Sue's OpenVMS bookmarks:
    http://eisner.encompasserve.org/~stu...bookmarks.html

+ Reply to Thread