(LONG) problem with multiple scsi devices on ESIX machine

[Jeff Carroll]

In article <1991Apr13.211855.4730 at unixland.uucp* bill at unixland.uucp (Bill Heiser) writes:
*
*I just got my Maxtor 8760S back from warranty repair, and have added it
*as follows:
*
*---on internal connect on Adaptec <----                --> on external connect
*CDC Wren HD --- Archive Tape --- Adaptec 1542b --- Toshiba  --- Maxtor
*  id 1                 id 4           id 7           id 2        id 0
*
*There is a termination at the CDC end and at the Maxtor end.
*
*I am not able to boot from my Esix floppies with this configuration.
*When I do, it gets to the secnd boot disk, then the machine hangs.
*There is a little disk activity for a minute after starting to read the
*second boot disk, but then nothing.  By removing either the Toshiba or
*Maxtor, and adjusting the termination/drive ids appropriately, everything
*is fine.    Also, with both drives in place, if I run SCSICNTL v5.5,
*it tells me "drive failed to respond to read capacity command" or some
*such thing.  The Toshiba is the culprit, it doesn't show up in the 
*"select device menu" in SCSICNTL.  Without the Maxtor in place, the
*Toshiba responds just fine!  The Maxtor responds either with or without
*the Toshiba in place.
*

I've seen Maxtor 8760S drives which behave very badly when put on the same
bus with an Exabyte drive - but they did so in such a way as to make it 
look like the Exabyte was the problem.

Our Maxtors would work either with the Exabyte present or without; but under 
heavy write-to-disk activity, the Maxtors would either spontaneously hang or
begin reformatting themselves if the Exabyte was on the bus.

I wouldn't automatically assume that the problem is with the Toshiba drive.

In our case the answer was to get separate SCSI controllers for disk and
tape. Not very attractive if you're on a budget...

*My first guess was a termination problem - but as shown above, I have
*termination at each end of the chain.  I'm a little fuzzy about whether
*each of the units should have parity-checking enabled/disabled, and
*termination power supplied to drive/bus, etc, ...  I've played with
*those a bit, and haven't seen a difference.  Maybe I haven't hit on 
*the right combination of configuration parameters???

I also would be very interested in any possible explanations of this problem.
All I can imagine is that we're dealing with a situation in which somebody's
compliance with the standard is teensily defective...


-- 
Jeff Carroll
carroll at ssc-vax.boeing.com

"Do you think I care? ... I have an infinite amount of money."	-Bill Gates