SUMMARY: SCSI drive errors on 4/370

From: Jeff Pack (
Date: Wed Apr 22 1992 - 16:43:54 CDT

Thanks again to sun-managers; you always come through when the vendors
don't. The original problem was SCSI bus resets on a 4/370 with 4 Maxtor
P-12 drives attached. Thanks to everyone. Suggestions came from:


>I don't know if this is your problem, but I had two Maxtor 1.0 GB
>drives fail in 4 months, then FINALLY the vendor informed me that the
>firmware on the disks was faulty. Seems the read-ahead cache causes
>various, intermittent errors/failures. I have some software that will
>determine the Rev # of the firmware and check to see if the read-ahead
>cache is turned on. Let me know if you would like me to email it to
>you. Unless you bought the drives VERY recently, I would bet this is
>your problem.

The drives are current; Danielle sent me the software and I was able to
determine that the firmware is current. But the software is very useful in
determining the status of the read-ahead cache. Thanks much, Danielle.



>Is there anything else on the SCSI bus besides the Maxtor drives? (Try
>a probe-scsi if the 4/370's ROM supports it.) I have seen stuff like
>this when trying to mix Adaptec- and Emulex-based shoeboxes on a sun-3
>under SunOS 3.5 .

Just the 4 Maxtor drives were on the bus, but we're getting warmer...



>anyway, the terminations should be at the ends of the bus.
>sm0 perhaps has one? phys sd2 should have it instead...

>sd3 should (and does, according to your note) have the other.

The wiring scheme on the 4/370 is strange; because of the peripherial
trays, it's hard to trace how the wiring goes. Sun verified that my
terminations were correct.



>The only odd thing I can see is that the snapshots refer to lun 1 all the
>time, and I take it that the drives are embedded SCSI (I don't know the
>drive designator off hand), so there are no LUN 1's at all, they're all 0.
>Is there anything running that's trying to access sd1/3/5 etc? I would be
>inclined to drop those lines from the kernel (you'll never use them) to
>avoid any abberant program trying to get at non-existent hardware.

Bingo! Although the only 4 devices on the bus were the Maxtor drives, the
kernel configuration defined additional devices on LUN 1. When I commented
out the additional devices, the symptoms disappeared.

So, the moral of the story is, if you don't need it, take it out of the

Thanks again, everyone.

Jeff Pack                         UUCP:  ...!uunet!!jfp
Idaho National Engineering Lab    Internet:
P. O. Box 1625 M.S. 2603          Phone:  (208) 526-0007
Idaho Falls, ID 83415             FAX:  (208) 526-9936

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:41 CDT