[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference vaxaxp::vmsnotes

Title:VAX and Alpha VMS
Notice:This is a new VMSnotes, please read note 2.1
Moderator:VAXAXP::BERNARDO
Created:Wed Jan 22 1997
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:703
Total number of notes:3722

439.0. "rules/precedence of reaccess to unloaded disks" by TAPE::SENEKER (Head banging causes brain mush) Wed Apr 09 1997 15:53

    Currently, the RWZ52 and RWZ53 optical disk drives perform different
    action regarding media ejection and spin up in response to a SCSI
    start/stop command. 
    
    I am looking for a description of design intent or precedence regarding
    this activity as it applies to OpenVMS system operation.  Hopefully
    this intent or precedence will determine which is correct, are both
    correct, are neither correct or if correct is even defined?
    
    Example:
    
    A removable disk is mounted in a drive, later the disk is dismounted
    with the /UNLOAD qualifier. Question, can or should this disk be
    allowed to be mounted without human intervention?
    
    Precedence based questions:
    
    o Does OpenVMS still support any removable magnetic disk drives?
      If so, what do these do in this case? 
      If multiple types of drives, do they all act the same?
    
    Design based questions:
    
    o Any OpenVMS security rules governing "re"access to "unloaded"
      disk media?
    
    Rob
T.RTitleUserPersonal
Name
DateLines
439.1mount and LoEjSTAR::EVERHARTWed Apr 09 1997 17:0734
    Well, OVMS still runs on VAXen, and those VAXen can still have
    ancient media like RM05, RM03, etc. on them...those are removable.
    Also many newer systems have floppy drives.
    
    Far as I can see there's no issue with load/unload of these; the
    volume label & VOLPRO priv control such things anyhow.
    
    Now, DKdriver's start/stop has never set the LoEj bit, nor has
    there been control for such things apart from someone using
    io$_diagnose to emit those functions in a SCSI start/stop packet.
    This has been known to cause grief with some 3rd party drives in
    jukeboxes; some of them require the bit, some require it to be off,
    and it would be a convenience to users if they could somehow set
    which behavior is used. (In fact something analogous can happen
    with tape drives too.) This is simply an artifact, again as far as I
    can tell, of history. VMS just doesn't attempt to get anything 
    specially loaded or unloaded; it will send SCSI start or stop to
    spin media up or down, but never an eject separately. This on the
    old removable pack devices was all that made sense anyway (images of
    RM05 packs flying around a room, destroying anything they collide
    with... :-) ). This area would make sense as yet another adjustable
    SCSI parameter, but the effort to build infrastructure to support
    such parameters is still in its infancy. Till a much later period,
    it can't be done.
    
    There is nothing policy wise to prevent a disk from being mounted
    without human intervention, and this has been the usual case for the
    old pack devices and for floppies, /unload or no. Should it be 
    difficult to extract the bloody thing from the drive, though,
    there is also nothing policy wise which would say that sending a
    scsi stop WITH the LoEj bit on would be wrong either. That it
    isn't done is not a conscious policy decision; it's just the way
    things happened.
    
439.2more informationTAPE::SENEKERHead banging causes brain mushWed Apr 09 1997 18:0838
    In the case of the RWZ5x drives the OSDS software is the same and
    the LoEj bit is always set on DISMOUNT/UNLOAD.  The variation is
    in the drive firmware control.  The RWZ52 will retract and reload
    the optical disk onto the spindle while the RWZ53 will not. Optical
    hardware and software engineering is trying to determine if their
    is a problem with these drives acting different other than the pain
    it causes some customers.

    Another concern is, "If it was done wrong for the RWZ52, should it
    be done wrong on the RWZ53 out of precedence?"  Also, what constitutes
    wrong?

    From a SCSI standpoint after both drives are in the "unload"ed state.
    The RWZ52 will complete a START UNIT command while the RWZ53 will
    return CHECK CONDITION and "media not present" sense data.

    RE: .1 - are you saying that OpenVMS currently allows removable media
    disk volumes to be "remounted" with no human intervention if the
    mounter's process has the proper priv's.  I am only interested in
    OpenVMS supported disk devices, not tapes, not optical disk, just the
    magnetic stuff that OpenVMS engineering would accept an IPMT case
    against.

    If OpenVMS doesn't have a rule or precedent then the optical group
    can try to just use the SCSI specification for guidance.  Even if
    OpenVMS has some ideas the optical group will have to create any
    possible fixes in the proper place so the drive can be used on non-VMS
    systems.

    Also, anybody else have input....
    Rob

    P.S. Our interpretation is that if the device is a desktop SCSI disk
	 that supports eject with the STOP UNIT command then a VMS "unloaded"
	 volume requires human intervention, to push in disk, before a
	 SCSI START UNIT command (or VMS MOUNT) will work.  We feel the
	 RWZ53 is working correctly and the RWZ52 is not.
439.3Consider floppySTAR::EVERHARTThu Apr 10 1997 10:467
    If you dismount an RM05, even with spindown, and then mount it, it
    spins up as I recall. A floppy certainly will spin up; the packack
    sends a SCSI start to the thing. No intervention is needed there
    since the thing's still in the drive.
    
    Since with RWZ53 the customer could always use dism/nounl, tho,
    I doubt the difference in behavior will bother anyone.
439.4I recall RM05 requiring manual restartEVMS::KILGALLENZK0 4x13, DTN 381-2879Thu Apr 10 1997 11:0322
>                       <<< Note 439.3 by STAR::EVERHART >>>

>     If you dismount an RM05, even with spindown, and then mount it, it
>     spins up as I recall.

My recollection is that once an RM05 was spun down, manual intervention
was required to spin it up again.  Perhaps there was a hardware change
to affect this behavior.

The introduction of DSA disks resulted in decreased reliability at the
site where I was consulting because disk-to-disk image backups were
used and it was possible with RM05s under command procedure control
to dismount the "old" disk, start running the "new" disk and have no
fear that an automatic reboot would mistakenly mount the "old" pack
(due to the crash happening just before the batch job modified the
list of which disks were on which drives).

As I recall we switched to a practice of operators interchanging
unit plugs as directed by the batch job, and of course that meant
we could use RA81s as well as RA60s, but it would have been much
better for a program to be able to prevent disk use until there
was human intervention.