Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!rice!sun-spots-request
From: mjacob@sun.com (Matt Jacob)
Newsgroups: comp.sys.sun
Subject: Re: Attaching non-sun SCSI drives to a SparcStation
Keywords: Hardware
Message-ID: <3852@brazos.Rice.edu>
Date: 8 Dec 89 07:33:28 GMT
Sender: root@rice.edu
Organization: Sun-Spots
Lines: 41
Approved: Sun-Spots@rice.edu
X-Refs:  Original: v8n205
X-Sun-Spots-Digest: Volume 8, Issue 220, message 2 of 19

In article <3246@brazos.Rice.edu> gert@fwi.uva.nl (Gert Poletiek) writes:
|X-Sun-Spots-Digest: Volume 8, Issue 205, message 5 of 19
|
|The problem we encountered is the following. During the last phase of
|formatting the disk (a CDC Wren IV) the disk label and backup label are
|written to the disk. The label is written correctly (since it is written
|onto one of the first blocks of the disk), but writing the backup label
|fails. We get a SCSI error back from the disk saying that the format
|program tried to write beyond the end of the disk.  We tried changing the
|format parameters in /etc/format.dat but that didn't help either. If
|nsyl=x, acyl=y and pcyl=x+y we always get 'No media error on cylinder x'
|nce the backup label is being written.
|
|We've experienced the same problem with other SCSI disks with an embedded
|controller. Sun disk with ESDI/SCSI convertor boards in them do not have
|this problem.
|
|I reported this problem to Sun in Holland and they told me: Sun ships its
|own CDC Wren disks.  They couldn't tell me what Sun had changed with
|respect to the 'normal' CDC Wrens. The only thing they could have changed
|is the ROM on the embedded SCSI controller board.
|
|Does anyone have a solution to this problem?

The CDC-WrenIV that I worked with for SS-1 was the model 94171-344. The
format.dat entry as supplied with SS-1 worked with that. Please remember
that since the CDC Wren-IV is a variable geometry drive, the ncyl, acyl
and pcyl values are constructs, so that if you change /etc/format.dat to
what you think *might* be a better value, you may be disappointed.

You aren't running with beta code are you? There was a bug that I think
that we fixed last spring where the sectorss/trk field was being set for
variable geometry drives (this had an unpleasant side effect of truncating
about 10mb off of the end of the disk).

Another thing- if the drive is seen w/o a label, on SS-1 it will print the
vendor string and the drive's reported capacity- this might be a clue as
to what is going on (like, the drive is reporting significantly less
capacity than it should have).

-matt jacob