SUMMARY: SCSI: "unexpected phase"

From: Jeff Boyd (BOYDJ@QUCDN.QueensU.CA)
Date: Sun Aug 15 1993 - 22:31:11 CDT


I received 3 replies; my thanks to the following for their
suggestions:

      geno@rtfm.mlb.fl.us (Gene Valicenti)
      aad@siemens.siemens.com (Anthony A. Datri)
      Pete Thurmes <pete@aps1.spa.umn.edu>

The initial assumption that installing the new drive caused the problem
has turned out to be wrong ... something has died in the SS1+ cabinet.
More than something actually, several things at once. I'm pricing new
machines now (ugh).

jeff

-------------------------------------------------------------------------

From: geno@rtfm.mlb.fl.us (Gene Valicenti)

Since you are still getting the error with the origional equipment then
I would suspect the scsi bus cable, a bent pin on the drive or a loose
power connection.

-------------------------------------------------------------------------

From: aad@siemens.siemens.com (Anthony A. Datri)

Check the cables -- I had things like this happen when one pin on one
of the high-density connectors got pushed in.

-------------------------------------------------------------------------

From: "Pete Thurmes" <pete@aps1.spa.umn.edu>

You state that there should be 6 devices found. Does that mean you
are trying to put 6 drives on one scsi bus?

Unless you have either very short ribbon cables and *very* short
external cables, or differential scsi, you may just have been lucky
to not have seen this problem before.

I have added drives to a bus and seen this "unexpected phase = 2"
error message. It means the scsi chain is too long and/or is
improperly terminated. Are you *sure* all your drives are
unterminated, or if internally terminated (one) that this drive
is at the end of the chain?

Suggestion for diagnosis: Remove a couple drives from the chain
and see if the problem goes away. If it does, try swapping those
removed drives for others in the chain. IF the problem reappears,
you probably have terminator SIPPs on one of the 2 drives
originally pulled. IF the problem does not reappear, it is no doubt
a scsi length problem. If the problem does not go away upon removing
drives from the end of the chain, you probably have an internally-
terminated drive in the chain or a bad terminator. I have had
"active" terminators (with the blinking LED) work where a passive
one would give SCSI errors, for a SCSI2 bus that was fairly long.

Be sure to remove drives from the end of the chain and terminate
the last drive. It is always a good idea to remove all SIPPs
from drives and terminate externally so that you don't inadvertently
put a terminated drive in the middle of a chain.

I know that SCSI termination is always blamed. But in my experience
it almost always *has* turned out to be the culprit. I have even
experienced the same "change-error-change back-error" cycle that
you have; it was still scsi chain errors. Sometimes these things will
work for months, and then you remove a cable to move a drive,
reconnect and you get errors. When a SCSI bus physical cabling
length gets longer than about 15 feet, you will start getting errors,
just sporadically if not *too* long, all the time if *really* too
long.

Remember that the SCSI driver on each drive adds its own effective
bus length, so physical cabling is only part of the equation.



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:06 CDT