I've received five responses from various sources, so far, and do not
except more; hence the summary. I've ordered them in, what I consider,
their order of significance. But do examine all.
Of the five, the first one below, from Michael Sullivan, has helped solve
my problem. Thanks, Michael for a most comprehensive answer. BTW, I have
asked Michael for the numbers of the OEM manuals as I believe this info
will permit me to get a set.
While awaiting responses I entered a call into Sun Canada's customer
service department and received much the same information that Michael
provided except without the detail. According to Sun there are no SunOS
patches that affect the drive; but Mory Katz' message forces me to
scrutenize the available patches much more closely. Hopefully a patch
will be found.
Thanks to all for the quick help,
==========================================
X-From: uunet!trdlnk!mike (Michael Sullivan)
X-Subject: Re: 1/2-inch SCSI problem
In article <5094@brchh104.bnr.ca> you write:
>I require assistance with one of Sun's 1.2-inch, front-loading tape drives
>(the HP unit) that will _NOT_ complete a multi-volume dump(8) or a sybase
>DUMP DATABASE. These multi-volume backups are performed on a 4/330 with
>a front-loading, 1/2 inch SCSI tape unit. The system is currently running
>SunOS4.0.3 with three 1G-IPI drives.
We have a similar set up here with a 4/380 and what sounds like the same
tape drive (I presume you meant "1/2-inch") except ours came directly from
HP. We are running SunOS4.1.1, but I don't think that is the problem
(although I should warn you that if you are planning on upgrading to
4.1.1, be sure to get patch 100250-01, which fixes an annoying bug with
the MTFSR and MTBSR ioctls that caused tar appends to silently fail).
>I've tried set size and other parameters to no avail because the dump
>eventually fails with a hard error either at the end of the first volume
>or the end of the second. The errors are generally the same. Likewise,
>using sybase's DUMP DATABASE command prduces similar results. Again
>either at the end of the first volume or the second.
.
>Only new tapes are used (Scotch usually) and the tape head is cleaned
>regularly so something else must be causing the failures.
We had a similar problem a few months ago. The tape drive had
particularly bad problems writing tape marks at the ends of volumes. Like
you, I tried a new tape, and cleaned the head regularly (or so I thought).
Finally, we had an HP serviceman come in, and he discovered that the head
was dirty after all. I had been cleaning the head with
trichlorotriflouroethane and swabs, as recommended in the manual. The
serviceman said to ignore the manual, and instead he gave me a bottle of
91% isoproyl alcohol and some cloth sleeves that fit over a plastic stick
that resembles a popsicle stick. I had always cleaned the head very
gently because I thought it was fairly delicate, but he said you have to
scrub it very hard (in the direction of tape motion) to get at oxide that
gets stuck inside microscopic scratches. He scrubbed it so hard that the
entire drive and the table on which it sits shook! The tape drive has
worked fine since then, so I presume his advice is correct.
You can check if the drive is the cause of the problem by running a few
tests on a scratch tape. Unfortunately, HP doesn't document the tests in
the User's Manual, but we have the OEM manual set, which lists all sorts
of fun tests.
The following sequence readily exhibited the problem.
1. Mount a scratch tape (write enabled).
2. Take the drive offline.
3. Press the OPTION key; you should see "TEST".
4. Press ENTER.
5. Select test 150 (write density id) with your favorite combination of the
NEXT, PREV, DENSITY, ONLINE and UNLOAD keys.
6. press ENTER
7. In response to the prompt "ONCE", press ENTER.
8. In response to the prompt "A 6250", press ENTER.
The drive will work for a bit, then hopefully display "PASS".
9. Press NEXT twice to get to test 152 (write tape mark).
10. Press PREV to see "LOOP".
11. Press "ENTER" to start the test.
The drive will then try to continuously write tape marks.
If it stops soon (seconds) and displays "FAIL", the drive has a problem.
You can also try running test number 1 (general checkout), which basically
runs every possible test, including writing and reading an entire scratch
tape in both densities. This takes a while!
==========================================
X-From: Tom Crummey <tom@sees.bangor.ac.uk>
X-Subject: Re: Problem with 1/2 SCSI Fron-load Tape unit
We had some similar problems with our front load 1/2 inch tape drive on a
4/390. Coupled with the spurious write errors, we found that the tape
drive would refuse to load tapes without 6 or 7 tries. The solution in our
case was a replacement drive.
Configuration is not a problem, the drive is auto everything, and if it is
the same as when you bought it, there should be no problem. It sounds as
if one of the write amplifiers for the heads has gone sick.
If you haven't got a support contract with Sun, the drive is made by
Hewlett Packard Type 88780B.
Hope this is of some help.
Tom Crummey
tom@uk.ac.bangor.sees
==========================================
X-From: jpl@allegra.att.com (John P. Linderman)
X-Subject: Re: Problem with 1/2 SCSI Fron-load Tape unit
We've seen similar problems on our 8mm SCSI tapes. In most cases, we've
had to swap out the drive, although cable swapping is worth a try. Just
yesterday, I replaced a drive that would die in the middle of any large
dump. As soon as I replaced the drive, dumps worked fine.
If you hear of a less draconian solution, I'd sure be interested. We have
quite a stack of failed drives waiting to be repaired.
==========================================
X-From: katz@rpal.rockwell.com (Morry Katz)
X-Subject: Problem with 1/2 SCSI Fron-load Tape unit
I'm having great difficulty performing multi-volume backups
on a 4/330 with a front-loading, 1/2 inch SCSI tape unit. The
system is currently running SunOS4.0.3 with three 1G-IPI drives.
The first is partitioned with root, /usr and /files, with the latter
partition holding a sybase database. The remaining two are recent
addition and hold data for the application. Further we must perform
backups correctly prior to making any attempt to upgrade the OS.
There are at least 3 patches for bugs in the SCSI device driver code for
OS4.1. I believe that the bugs these fix were also present in 4.0.3, so
you probabaly need to install some or all of the 4.0.3 patches for the
SCSI tape devies in order to get a good dump.
==========================================
X-From: Steve Simmons <scs@wotan.iti.org>
X-Subject: Re: 1/2-inch SCSI problem
We saw the same problem once with a defective head on the tape drive. The
head had literally worn out.
Also, odd read/write errors can occur if the mounting hubs wear. When the
tape goes thru a stop-start cycle, it can slip slightly on the hubs. I've
seen this happen on older, heavily used Fujitsu drives (the ones sun
shipped with the 3/160 and 3/260). I've no idea what the HP hubs are
like, but the Fuji had 3 little wheels that pushed against the inside of
the reel to lock it in place. Over time the wheels would chip, reducing
the diameter and thereby reducing the locking force.
==========================================
X-From: uunet!tc.fluke.COM!pwl (Paul Lutt)
X-Subject: Re: Problem with 1/2 SCSI Fron-load Tape unit
We've been fighting problems with the HP 1/2" tape drives for the last
year. There are ECOs for the tape drives. SunOS 4.0.3 seems to have lots
of problems with these drives. Sun seems to have fixed alot of the
problems in the SunOS 4.1.1 release. Contact your local Sun office and
complain, complain, complain. Upgrading to SunOS 4.1.1 would help. Good
luck.
Paul Lutt
==========================================
X-From: uunet!force-c!doc!kcc (Ken Coley)
X-Subject: Re: Problem with 1/2 SCSI Fron-load Tape unit
What type of tape drive does your power up message say you have?
- Archive 2150s?
- Archive 2060s?
What type of tapes are you using?
Ken Coley
Force Computers
==========================================
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:19 CDT