Spinning down an old disk array - VMS

This is a discussion on Spinning down an old disk array - VMS ; A customer is planning on doing some maintenance at their data centre. As part of the maintenance their HSZ80 and HSJ50 disk sub-systems will have their power cut off. Since most of these disks have been constantly spinning for the ...

+ Reply to Thread
Results 1 to 10 of 10

Thread: Spinning down an old disk array

  1. Spinning down an old disk array

    A customer is planning on doing some maintenance at their data centre.
    As part of the maintenance their HSZ80 and HSJ50 disk sub-systems will
    have their power cut off. Since most of these disks have been
    constantly spinning for the past 8 or 9 years the customer is
    concerned about the disks spinning again after power is restored.

    Most disks are DR-RZ1FC-VW and some are RZ29.

    Some people here feel that as long as the power is off for only 10 or
    15 minutes that the disks should spin up again after power is
    restored. Some people here feel that even if the power is off for a
    few seconds that we risk having disks not spin again.

    Does anyone have any experience with turning off the power on disks
    that have been running for years? What percentage of disks should we
    expect to have fail after;
    - a few seconds
    - a few minutes
    - 10 minutes
    - 15 minutes


    Peter

  2. Re: Spinning down an old disk array

    On Sep 10, 10:25*am, Peter Weaver
    wrote:
    > Does anyone have any experience with turning off the power on disks
    > that have been running for years? What percentage of disks should we
    > expect to have fail after; ...


    Yes to the first part. I wouldn't say frequently, but certainly I
    have turned off complete disk arrays for maintenance and then powered
    them right back up again without incident.

    Too random on the second part. In my experience: none failed. In
    fact, I'd say I've had better experience with the older 5400rpm drives
    than newer 10k or 15k drives.

  3. Re: Spinning down an old disk array

    On Sep 10, 9:25 am, Peter Weaver wrote:
    > A customer is planning on doing some maintenance at their data centre.
    > As part of the maintenance their HSZ80 and HSJ50 disk sub-systems will
    > have their power cut off. Since most of these disks have been
    > constantly spinning for the past 8 or 9 years the customer is
    > concerned about the disks spinning again after power is restored.
    >
    > Most disks are DR-RZ1FC-VW and some are RZ29.
    >
    > Some people here feel that as long as the power is off for only 10 or
    > 15 minutes that the disks should spin up again after power is
    > restored. Some people here feel that even if the power is off for a
    > few seconds that we risk having disks not spin again.
    >
    > Does anyone have any experience with turning off the power on disks
    > that have been running for years? What percentage of disks should we
    > expect to have fail after;
    > - a few seconds
    > - a few minutes
    > - 10 minutes
    > - 15 minutes
    >
    > Peter


    Peter,

    The original post does not indicate how many of these drives are in
    stripes, mirrors, and other flavors of RAID. For certain, particularly
    because the term "maintenance" includes much real estate (including
    power and water), I would recommend that backups be up-to-date and off-
    site during the "maintenance".

    That said, I have not seen particularly bad experiences caused by a
    single power down. In my experience, most of the interesting problems
    come on sites where power-up/power-down is a chronic cycle, and the
    cumulative wear and tear does cause failures.

    It also has a tendency to uncover out-of-date batteries in various
    devices. Perhaps one of the more overlooked checklist items is making
    sure that systems and controllers have up to date NVRAM and other
    batteries. Spare batteries would not be a bad idea, as is using the
    opportunity to change batteries for fresh ones while the systems are
    powered down.

    - Bob Gezelter, http://www.rlgsc.com

  4. Re: Spinning down an old disk array

    Peter Weaver wrote:
    > A customer is planning on doing some maintenance at their data centre.
    > As part of the maintenance their HSZ80 and HSJ50 disk sub-systems will
    > have their power cut off. Since most of these disks have been
    > constantly spinning for the past 8 or 9 years the customer is
    > concerned about the disks spinning again after power is restored.
    >
    > Most disks are DR-RZ1FC-VW and some are RZ29.
    >
    > Some people here feel that as long as the power is off for only 10 or
    > 15 minutes that the disks should spin up again after power is
    > restored. Some people here feel that even if the power is off for a
    > few seconds that we risk having disks not spin again.
    >
    > Does anyone have any experience with turning off the power on disks
    > that have been running for years? What percentage of disks should we
    > expect to have fail after;
    > - a few seconds
    > - a few minutes
    > - 10 minutes
    > - 15 minutes
    >
    >
    > Peter


    Sooner or later EVERY disk will fail! People use BACKUP to ensure that
    no data is lost. Various forms of RAID are used to ensure that access
    to data is not lost.

    Ideally, you should have spares on hand for each make and model of disk
    drive in use. It's easy if all your disks are StorageWorks; just pop a
    failed drive out of its socket and plug in a new one.

    If anything fails at power on, I would expect it to happen within the
    first sixty seconds or less.



  5. Re: Spinning down an old disk array

    I have found that if they are off long enough to cool down then more
    of them fail.

    To add to what Bob said - Do ensure you know the settings stored in
    NVRAM so when they fail you can replace the battery and re-enter the
    old settings.

    Do have a set of spare disks and set expectations that there will be
    some failures.


  6. Re: Spinning down an old disk array


    "Peter Weaver" wrote in message
    news:e97191b9-77c7-4620-8f85-276279907ba5@34g2000hsh.googlegroups.com...

    > Does anyone have any experience with turning off the power on disks
    > that have been running for years? What percentage of disks should we
    > expect to have fail after;
    > - a few seconds
    > - a few minutes
    > - 10 minutes
    > - 15 minutes


    I have 4 pair of HSZ70s with various blue bricks inside (9GB, 18GB, 36GB, 96
    drives total). Power cycles (minutes to days) haven't been a problem.
    Still keep current backups and some drives are mirrored on the controller.

    I recently had to move them to another site (pushed down the hallway, into
    an elevator, onto a truck, driven lets say 48 miles to another state, back
    off the truck and into their new location). 96 drives in total, 2 died at
    the other end (1 9GB drive and 1 18GB drive). We restored the backups from
    tape.

    I did try to "thermal cycle" the bad drives that wouldn't spin/mount. I
    sealed them into a plastic bag and put them in the freezer overnight [it has
    worked for me in the past]. One actually tried to spin up for only for
    about 30 seconds. Not long enough to mount/copy any files.

    I'll guess the drives died not from turning them off, but the vibration of
    the move.

    John



  7. Re: Spinning down an old disk array

    On 10 sep, 16:25, Peter Weaver wrote:
    > A customer is planning on doing some maintenance at their data centre.
    > As part of the maintenance their HSZ80 and HSJ50 disk sub-systems will
    > have their power cut off. Since most of these disks have been
    > constantly spinning for the past 8 or 9 years the customer is
    > concerned about the disks spinning again after power is restored.
    >
    > Most disks are DR-RZ1FC-VW and some are RZ29.
    >
    > Some people here feel that as long as the power is off for only 10 or
    > 15 minutes that the disks should spin up again after power is
    > restored. Some people here feel that even if the power is off for a
    > few seconds that we risk having disks not spin again.
    >
    > Does anyone have any experience with turning off the power on disks
    > that have been running for years? What percentage of disks should we
    > expect to have fail after;
    > * * *- a few seconds
    > * * *- a few minutes
    > * * *- 10 minutes
    > * * *- 15 minutes
    >
    > Peter


    Peter, I don't think there's a definitive answer to your question.
    Some of the possible RAID configurations may offer recovery
    possibilities should one drive fail. Other than that I'd expect full
    backups of all drives that are part of this maintenance project.
    At home I run systems that are 10 to 18 years old. Most of these
    systems are powered off most of the time. I haven't had a SCSI disk
    fail, yet. But these disks are 5400 and 7200 rpm disks with capacities
    of 2, 4 and 9 GB. Furthermore, all disks run in a horizontal position,
    so the bearings are not affected by gravity.
    Hans

  8. Re: Spinning down an old disk array

    FWIW:

    Back when I ran my all mighty Microvax II on RD54, I was concerned about
    its age. But the RD54 never quit. But I was given an RD53 and I
    connected it and it worked. It started to log a few errors. So I
    shutdown the system to check connectiosn etc, and when I restarted it,
    the RD53 started to spin up, but never accelereated to normal speed.
    (the way I saw it, it was like when a jet aircraft starts an engine, the
    engine starts to spin from the starter, but it is only once they apply
    the fuel and ignite it that it *really* starts to spin up. That disk was
    like in "starter" mode and didn't ignite...


    But for a newer generation SCSI disk (250meg), it had come off an old
    mac. It started to log errors while running. Not sure if it was the
    motor failing or the heads/surface.


    In the case of a disk array, my concern would be with regards to power.
    If you are goin to power up the whole thing, does it have logic to
    sequentially power up drives, or would they all get their power at the
    same time ? Are they set to spin up automatically, or await a spin up
    command from the SCSI bus ?

    If you could power up the drives one after the other, it might be better
    because of less strain on the power supply and each drive would then
    not see as big of a power spike. You could also more easily see which
    drive is spinning up and which isn't.

    If they have been running for that long without many interruptions and
    in good environment (temperature not too hot) I woudln't be *too*
    worried about then not spinning up.

    Would starting those disks while still warm be easier ? I am thinking if
    there is any oil in the motor, bearings whatever, old oil might seize if
    allowed to cool down and offer more resistance to a power up, whereas if
    still warm, it would still be fluid ? Would that be a concern, or is
    that totally offbase ?

  9. Re: Spinning down an old disk array

    John Reagan wrote:
    > "Peter Weaver" wrote in message
    > news:e97191b9-77c7-4620-8f85-276279907ba5@34g2000hsh.googlegroups.com...
    >
    >> Does anyone have any experience with turning off the power on disks
    >> that have been running for years? What percentage of disks should we
    >> expect to have fail after;
    >> - a few seconds
    >> - a few minutes
    >> - 10 minutes
    >> - 15 minutes

    >
    > I have 4 pair of HSZ70s with various blue bricks inside (9GB, 18GB, 36GB, 96
    > drives total). Power cycles (minutes to days) haven't been a problem.
    > Still keep current backups and some drives are mirrored on the controller.
    >
    > I recently had to move them to another site (pushed down the hallway, into
    > an elevator, onto a truck, driven lets say 48 miles to another state, back
    > off the truck and into their new location). 96 drives in total, 2 died at
    > the other end (1 9GB drive and 1 18GB drive). We restored the backups from
    > tape.
    >
    > I did try to "thermal cycle" the bad drives that wouldn't spin/mount. I
    > sealed them into a plastic bag and put them in the freezer overnight [it has
    > worked for me in the past]. One actually tried to spin up for only for
    > about 30 seconds. Not long enough to mount/copy any files.
    >
    > I'll guess the drives died not from turning them off, but the vibration of
    > the move.
    >
    > John
    >
    >


    If you had packed the drives in containers similar to those they were
    shipped to you in, they all would probably survived the trip. That
    sponge rubber cradle would absorb a lot of shock and vibration!


  10. Re: Spinning down an old disk array


    "JF Mezei" schreef in bericht
    news:48cac674$0$9668$c3e8da3@news.astraweb.com...
    > FWIW:
    >
    > Back when I ran my all mighty Microvax II on RD54, I was concerned about
    > its age. But the RD54 never quit. But I was given an RD53 and I
    > connected it and it worked. It started to log a few errors. So I
    > shutdown the system to check connectiosn etc, and when I restarted it,
    > the RD53 started to spin up, but never accelereated to normal speed.
    > (the way I saw it, it was like when a jet aircraft starts an engine, the
    > engine starts to spin from the starter, but it is only once they apply
    > the fuel and ignite it that it *really* starts to spin up. That disk was
    > like in "starter" mode and didn't ignite...
    >

    JF, the RD54 and RD53 (Maxtor disks IIRC) and both models have short
    lifespans, say 12 to 14 years when used (switched on) most of the time.
    Possibly Maxtor designed these disks for a much shorter MTBF, no idea. But
    SCSI devices tend to live a lot longer.

    >
    > But for a newer generation SCSI disk (250meg), it had come off an old
    > mac. It started to log errors while running. Not sure if it was the
    > motor failing or the heads/surface.
    >

    Motor and bearing failures are noticable, sudden loud noise usually hight
    pitched is a bad sign.
    Though I have a couple of Barracuda's that scream as if tortured but
    otherwise perform well and errorfree.

    >
    > In the case of a disk array, my concern would be with regards to power.
    > If you are goin to power up the whole thing, does it have logic to
    > sequentially power up drives, or would they all get their power at the
    > same time ? Are they set to spin up automatically, or await a spin up
    > command from the SCSI bus ?
    >

    You're referring to power surges at startup time, right? I'm sure the effect
    is there but these disks are no RA81's :-)
    Besides those disks were wired to start up sequentially otherwise they'd
    blow a fuse.

    > If you could power up the drives one after the other, it might be better
    > because of less strain on the power supply and each drive would then
    > not see as big of a power spike. You could also more easily see which
    > drive is spinning up and which isn't.
    >

    You've got a point here, unless the diskbays have dual power supplies.

    > If they have been running for that long without many interruptions and
    > in good environment (temperature not too hot) I woudln't be *too*
    > worried about then not spinning up.
    >
    > Would starting those disks while still warm be easier ? I am thinking if
    > there is any oil in the motor, bearings whatever, old oil might seize if
    > allowed to cool down and offer more resistance to a power up, whereas if
    > still warm, it would still be fluid ? Would that be a concern, or is
    > that totally offbase ?


    The heads are parked in a landing zone. Which may be a very rarely used
    location. Once the heads are there they may get stuck in that position. The
    motor bearing(s) would worry me. When the disk bearing was perpendicular to
    gravity (disk on its side) the low side of the bearing would wear out a
    little more that the topside. When shutdown and power off, the bearing cools
    down and the metal will shrink (very little but will be there). The spindle
    might get stuck in the worn out bearing and never spin up again. After a
    short while I guess the motor will burn out.
    Hans



+ Reply to Thread