Greetings once again:
The IDS (Irritated Disk Syndrome) problem, namely excessive SCSI
bus resets and warning messages at boot time, was apparently the
result of the drive's lack of tagged queueing support. As suggested
by Endre Polgar, adding the line
set scsi_options=0x178
to /etc/system and rebooting cleared everything up perfectly. All of
the resets and warnings are gone, and the drive continues to hum
along at 10 MB/sec synchronous.
When we started running Solaris on this machine a couple of years
ago, the solution offered by Sun's tech support was to switch to
async mode. Maybe they could learn a thing or two from this list.
;-)
Many thanks to Endre Polgar and Kevin Sheehan for their kind help.
The original request and their replys (in full) are appended.
Thanks also to anyone else whose reply is received after this
summary is posted.
=D.C. Williams <dcwill@ee.unr.edu>
=============================================================
The original message:
We're running a Toshiba MK-538FB drive on a system running Solaris 2.4
(recently upgraded from 2.2). Under 2.2, the drive was running in
SCSI asynchronous mode. A complete install (rather than an upgrade)
to 2.4 was performed, and this drive is now running in synchronous
mode. Post-boot operation seems normal in all respects, but a number
of warning messages are displayed during the boot sequence. We see
things like "MESSAGE OUT phase parity error", "Target 2 rejects our
message 'SIMPLE QUE TAG'", and several timeout messages. This is followed
by a SCSI bus reset and a reminder that these messages are informational.
After these hiccups, the system and the drive seem to run flawlessly.
scsiinfo confirms 10 MB/sec sync operation, and I think that's what
we'd like to keep (if possible). However, I would like to clean up
the boot process to avert the various warnings and bus resets.
I would greatly appreciate any suggestions, comments, or pointers
on how we might be able to correct this situation. Many TIA.
============================================================
>From epolgar@petra.icon.hu Tue Mar 7 21:48:32 1995
Return-Path: epolgar@petra.icon.hu
From: "Endre Polgar Technical Support Engineer ICON Ltd." <epolgar@petra.icon.hu>
Date: Wed, 8 Mar 1995 06:44:57 +0100
Subject: Re: SCSI sync/async operation
Hello,
I'm not sure (don't know this Toshiba drive) but I think it doesn't support
tagged queuing. You can turn off tagged queuing by appending the following line
to /etc/system and then rebooting the machine:
set scsi_options=0x178
It's worth giving it a try since it's harmless. Other values are:
0x58 for slow async mode, and
0x1f8 for fast sync mode w/ tagged queuing (the system default).
Hope this helps.
Endre
'''
(o o)
-----------------------o00--(_)--00o------------------------
| Endre Polgar (Mr.) | ICON Ltd. |
| technical support engineer | Miklos ter 2. |
| epolgar@icon.hu | Budapest |
| Tel.: +36-1-250-0440 | 1035 |
| Fax : +36-1-250-0395 | Hungary |
------------------------------------------------------------
=====================================================================
>From kevin@uniq.com.au Tue Mar 7 17:54:34 1995
Return-Path: kevin@uniq.com.au
From: Kevin.Sheehan@uniq.com.au (Kevin Sheehan {Consulting Poster Child})
Date: Wed, 8 Mar 1995 11:53:11 EST
Subject: Re: SCSI sync/async operation
[ Regarding "SCSI sync/async operation", dcwill@ee.unr.edu writes on Mar 7: ]
> Greetings:
> This was previously posted to Usenet without any response.
> We're running a Toshiba MK-538FB drive on a system running Solaris 2.4
> (recently upgraded from 2.2). Under 2.2, the drive was running in
> SCSI asynchronous mode. A complete install (rather than an upgrade)
> to 2.4 was performed, and this drive is now running in synchronous
> mode. Post-boot operation seems normal in all respects, but a number
> of warning messages are displayed during the boot sequence. We see
> things like "MESSAGE OUT phase parity error", "Target 2 rejects our
> message 'SIMPLE QUE TAG'", and several timeout messages. This is followed
> by a SCSI bus reset and a reminder that these messages are informational.
> After these hiccups, the system and the drive seem to run flawlessly.
> scsiinfo confirms 10 MB/sec sync operation, and I think that's what
> we'd like to keep (if possible). However, I would like to clean up
> the boot process to avert the various warnings and bus resets.
> I would greatly appreciate any suggestions, comments, or pointers
> on how we might be able to correct this situation. Many TIA.
Remember that when you use third party disks, you are hoping they get
everything right. In many cases, this isn't do. Caching control was
one that took a lot of vendors some time to get right.
The new place to fumble is tagged queueing. Many vendors don't get it
quite right. Try turning off tagged queueing (by setting scsi_options
in your /etc/system file - /usr/include/sys/scsi/impl/autoconf.h has
all the relevant bits) and see if that makes a difference.
If not, check your cabling - long cables are Bad. Too many devices with
long cables is Bad. Termination has to be right. Mismatched impedance
cables and connectors is not so good.
l & h,
kev
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:18 CDT