Hi,
I've been asked to help a group that are experiencing a problem
with a DEC ALPHA 2000, but I'm not sure what is wrong myself.
Here is the setup;
DEC AXP 2000 OSF/3.0, FIRMWARE 1.4 (I've been told)
1 RZ26 1Gyte internel disk.
Built in CD-ROM.
1 of three SEAGATE HAWK (ST15230N or ST32430N)
Problem
=======
When using external disk (internal devices work fine) the following occurs;
1) System's SCSI subsystem "locks up", kernel and all applications
continues to run but can't access any disks and then hang when they need
too.
From the command monitor;
PROM: >>> show dev
then fails on SCSI devices.
2) Power cycling disk unit often un-hangs the system and it then continues.
(If this fails, then it is necessary to power cycle the system itself.)
3) System does not report errors, this includes configured
syslogd, binlogd, scu and PROM test routines.
4) Exercise tools were not available.
Lack of time prevented setting up and running kernel debugging mode.
But I will do these is anybody thinks it will help.
5) Fault occurs with near identical behaviour on the three disk drives
tested.
Unfortunately we only had three Seagate Hawks to test.
2 x Seagate ST15230N
1 x Seagate ST32430N
Disks work fine on Alpha 3000,Silicon Graphics and Sun.
The fault also occured when using a HP DAT drive which was in the
same enclosure as one of the ST15230Ns.
6) Fault occurs unpredictably, but always whist external devices are in use.
(typically large file transfers, but also during small file transfers
as well.)
7) Both configured and generic kernels were tried.
8) I used two different tested SCSI cables and two active terminators.
Four possible faults stick in the mind;
1) Faulty SCSI card.
2) Bug in Firmware and/or operating system on Alpha.
3) Bug or incompatibility between AXP-2000 and Seagate Hawk disk drives
(firmware).
4) Fault on Alpha.
We are also contacting DEC support and getting them to upgrade FIRMWARE and OS,
but I wondered if anybody has seen a similar problem. (Disk is third party.)
I have found details of a similar fault occurring when IBM disk drives
were used on HP 9000/700 UNIX workstations. After a lot of investigation,
it turned out the fault resided in incompatible firmware on the IBM disk
drives and the fault was resolved by upgrading the firmware ROM chips on
each disk drive. Hence the question concerning the AXP 2000.
Thanks,
Tristan
Tristan Green, CCLRC Daresbury Laboratory
Received on Tue Mar 12 1996 - 19:30:52 NZDT