Re: Adaptec 2940UW and Aborted Reset.

Chuck Mead (chuck@moongroup.com)
Sat, 26 Sep 1998 19:45:12 -0400 (EDT)


On Sun, 27 Sep 1998, Matthew Hawkins wrote:

> On Sat, 26 Sep 1998, Chuck Mead wrote:
>
> > I am curious about the 2940UW drivers for Linux and am also wondering
> > if the following is a normal condition and if not... how to fix it:
> > (I have marked the area of concern with comments)
>
> First of all, set up your termination in the SCSI BIOS properly:
> (if for no other reason than to cut down on this crap at bootup ;)
>
> > Sep 25 14:01:11 server kernel: aic7xxx: Warning - detected
> > auto-termination on controller:
> > Sep 25 14:01:11 server kernel: aic7xxx: <Adaptec AHA-294X Ultra SCSI
> > host adapter> at PCI 9/0
> > Sep 25 14:01:11 server kernel: aic7xxx: Please verify driver detected
> > settings are correct.
> > Sep 25 14:01:11 server kernel: aic7xxx: If not, then please properly
> > set the device termination
> > Sep 25 14:01:11 server kernel: aic7xxx: in the Adaptec SCSI BIOS by
> > hitting CTRL-A when prompted
> > Sep 25 14:01:11 server kernel: aic7xxx: during machine bootup.
> > Sep 25 14:01:11 server kernel: aic7xxx: Cables present (Int-50 NO,
> > Int-68 YES, Ext-68 NO)
> > Sep 25 14:01:11 server kernel: aic7xxx: Termination (Low ON, High ON)

So auto termination is bad? I did some reading which indicated it
was okay to do this on Linux. No?

> > ***** Here is the area of my concern!
> > \
> > Sep 25 14:01:11 server kernel: scsi0 channel 0 : resetting for second
> > half of retries.
> > Sep 25 14:01:11 server kernel: SCSI bus is being reset for host 0
> > channel 0.
> > Sep 25 14:01:11 server kernel: scsi : aborting command due to timeout
> > : pid 18, scsi0, channel 0,
> > id 0, lun 0 Read Capacity 00 00 00 00 00 00 00 00 00
> > Sep 25 14:01:11 server kernel: (scsi0:0:0:0) Using wide (16 bit)
> > transfers.
> > Sep 25 14:01:11 server kernel: (scsi0:0:0:0) Synchronous at 10.0MHz,
> > offset 8.
> >
> > ***** Shouldn't this be 20.0MHz on a UW?
>
> Yes. Or 40Mb/s (the beta drivers now display transfer rate).
>
> > The drives themselves are 4.3GB IBM UW's... there are two of them but
> > the 2nd drive is not in use at this time. My intent is to use it
> > as an on-line backup but I haven't configured that yet.
>
> Use the md driver to do RAID1 - much easier than trying to dick about with
> backup cron jobs etc.

I really don't want raid... I just want a reliable back up that I can
plug in and restore service quicker than a tape!

> > What I am concerned about is the timeout and also the 10.0MHz result
> > (I thought it should be 20). I'm not sure if I have a problem
> > here... everything works and the box has been in production for more
> > than 60 days.
>
> First of all, go into the SCSI BIOS and set the SCSI id's which your drives
> are on to 40Mb/s. Then set the manual termination properly, and save the
> settings.

I can do this but will it break the running install...? If there's a
chance it will I can't do that now or at least until I have a
reliable external backup.

> > One additional piece of information is that when the initial install
> > was done I had a horrible time getting the RH install to see the
> > 2940UW. I could not do a simple default install and it finally
> > worked when I installed from expert mode.
<snip>
> I also had lots of scsi-related problems until plugging
> in one of Doug Ledford's beta scsi drivers (v5.1.0p7). Not only did the
> problems disappear, but both drives were recognised as having 40Mb/s
> transfer rate (as they do).

Where do I get doug's driver from?

> Squid still segfaults on the box every few days, but that's better than
> every few hours like it used to, with kernel oops' taking the whole server
> down. I still haven't been able to track down why, perhaps it's having
> two 128Mb SDRAM chips on a TX board (it's not processor, cache or swap).
>

Welp... one thing that I can say is that in spite of my concerns with
the SCSI bus and the logs mine has been running steady in production
for over 60 days now and there hasn't been an oops or anything else
(not even a core file)!

Thanx for your reply!
--
Chuck Mead, CEO - Moongroup Consulting, Inc.
<chuck@moongroup.com>
http://www.moongroup.com/
http://www.moongroup.com/unix/

Sometimes I wonder if I'm in my right mind. Then it passes off and I'm
as intelligent as ever.
-- Samuel Beckett, "Endgame"

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/