Flaky SCSI bus

From: Steve Timm <timm_at_eve.phy.albany.edu>
Date: Thu, 21 Aug 1997 22:51:57 -0400 (EDT)

Hi, managers....

I have a DEC3000-700 with three disks, all hooked to the internal
SCSI bus. The system disk is an RZ26 (just newly replaced) and
then I have two X3653 HP drives in the system as well. These drives
run rather hot and so we have jumpered them to only spin up when accessed.

We had been getting periodic SCSI bus resets for months, almost ever
since we installed the HP drives, but recently we crashed and couldn't
get back up due to multiple SCSI errors. At the time all the error
messages had as their target the system disk, so we've now replaced
the system disk.

Once the system disk was replaced, we still found ourselves unable to boot
due to many cam_scsi errors in the reboot process, even when the system
disk was the only disk in the system. After plugging and replugging
a few times, eventually we came up with a situation where first one
drive at a time, then two drives, and finally 3 drives were readable.
But it seems a rather probabilistic way to go and I have no guarantee
that the system will boot again when the power goes out.

Any questions here what the likely culprit may be? Is it a flaky SCSI
controller or rather just that we are pushing the bus to its limits,
or does the slow spinup of the drives exacerbate this problem?

Steve
Received on Fri Aug 22 1997 - 05:07:19 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:36 NZDT