Conference vaxaxp::vmsnotes

133.0. "RW500 and OpenVMS V6.x or later ???" by FIREBL::LEEDS (From VAXinated to Alphaholic) Mon Feb 03 1997 12:53

Got the attached from a customer. Any suggestions ??

Regards,

Arlan

Our story thus far:

I have a DEC RW 500 optical disk jukebox in my office.  The optical
drive inside has the annoying habit of reporting itself as a SCSI
device type "OPT" (for optical) instead of "DIR" (for direct-access).
This makes the standard SCSI disk driver, which is looking for "DIR"
type devices,  reluctant to talk to the optical drive.

Under OpenVMS VAX, this drive worked just fine under VMS 6.0; I was able to
use the jukebox by commanding the robot with my own software and using
VMS to read/write the optical drive via the standard SCSI driver.

Under VMS 6.2, this failed.  All operations to the optical drive
failed with the most popular error being "medium is offline".  My
information from DEC telephone support is that this was deliberate;
the VMS SCSI driver was made to become a real stickler about the SCSI-2
standard.  The SCSI disk driver recognized the device, but would not
talk to it.

Under VMS 7.0, the situation degraded even further, to the point that
the SCSI disk driver failed to recognize that the drive even existed.

So the upshot is that, by the simple act of upgrading VMS, DEC's own
peripheral stopped working, and furthermore, that DEC broke it on pur-
pose, albeit indirectly.

Now I know that DEC has some snazzy jukebox management software that
will turn the jukebox into a single humongous file system and use the
robot to bake bread when it isn't busy changing disks and all kinds of
other fancy stuff.  I don't need it, and I don't want it.  I just want
for the SCSI disk driver should talk to the optical disk.

Fortunately for me, personally, Ultrix handles this quite nicely.
Through scu, you can *tell* Ultrix to pretend that the SCSI drive
which reported itself as type "OPT" is really type "DIR".  The bog-
standard SCSI disk driver is then more than happy to talk to the
drive, and the jukebox works just fine under Ultrix.

This same procedure almost works under OSF/1; misfortunately, scu
gives some utterly irrelevant error message and refuses to change
its idea about the drive type.

But that's another story; my primary concern is about VMS.  I'd
*really* like to get a VMS SCSI disk driver for VMS 6.2 and 7.0
(and *beyond*) that will talk to the optical disk drive in the
same way that VMS 6.0 did.

Anything you might be able to do would be greatly appreciated.

Thanks.

:But that's another story; my primary concern is about VMS.  I'd
:*really* like to get a VMS SCSI disk driver for VMS 6.2 and 7.0
:(and *beyond*) that will talk to the optical disk drive in the
:same way that VMS 6.0 did.

   The optical notes conference is ELWOOD::OPTICAL.

   Device configuration changed starting in OpenVMS V7.1 -- see the
   release notes for details.

   See if the mode pages on the device can be altered -- you'll need
   to check with the optical folks here -- to report direct-access,
   and/or see if there is a device driver available for this device.

I was just coming back here to say "also posted in ELWOOD::OPTICAL" when I 
saw .1 --- I'll continue to follow it there...

Arlan

    A bunch was done to dkdriver for 7.1 to get it to work with opticals,
    but mind there are none to test on to speak of (well, there are
    a couple & it works there). I'd expect that code will continue
    to be there, but it is NOT part of qual. The 6.2 and 7.0 drivers
    did not have this stuff. The common boot driver was also modified
    to understand optical as an "ok" device type to use...and VMS
    has booted off optical disks. But agtain this is stuff that
    is not officially supported.
    glenn
    

    Glenn, does DKDRIVER still require a Geometry page for MO's? The
    Digital drives have them (because we added them, to work around the
    DKDRIVER need) but most MO's don't... 
    
    Chris
    

    DKdriver will generate a geometry if it doesn't find one now. As long
    as the device size in blocks is present, you'll get a geometry that
    gets you access to nearly all of the device, cut from whole cloth if
    necessary. This is true for MOs or normal disks or whatever. This is
    so in the 7.1 flavor. I have some experimental drivers for 6.2 and
    7.0 alpha with the fix also if needed, but no guarantees or support for
    them...if you're internal ya gets the code and takes yer chances :-) .
    
    Glenn
    

    We did have an earlier problem with one of the pseudo-geometries that
    OSDS/OSMS generated when it could not get to all the blocks on the
    disk.. We had to change it to allow access to all blocks, after some
    complaints....  Care is needed here.
    
    
    Chris