In response to my earlier posting about CAM SCSI errors on two machines that had
been running successfully for 14-24 months, I wish to thank the following people
for responses: tpb_at_zk3.dec.com, seb_at_lns62.lns.cornell.edu, ebishop_at_vt.edu,
alan_at_nabeth.cxo.dec.com, kchiles_at_hccsf.com, chris_at_lagoon.meo.dec.com,
Richard_Fleming_at_ccmsmtpgw.gam.com. (My apologies for the delay in this summary
posting, but I was out for nearly 2 weeks.)
I had initially sent only a summary uerf output - my apologies, as I was
reminded by several to run uerf with the -o switch to get the full output.
In both instances, the disks had to be replaced. In one instance, a later
firmware revision level is what we needed; in the other, it was just defective
(didn't power up upon reboot necessarily). When installing disks, ask field
service to tell you what the firmware revision is - that way you can ask
specifically for the latest rev, or at least for any Emergency Change Orders
(ECOs). DEC field service tried to install new disks twice into one machine,
and both had the same firmware revision and problems as our original disk.
(Note: for smaller sites at least, DEC doesn't send this type of firmware
upgrade, and field service doesn't track it.)
A summary of a response from seb_at_lns62.lns.cornell.edu may be helpful as well:
"Check that a SCSI terminator is plugged into the connector on the back of the
CPU (for the internal SCSI disks). On 3000-600, the firmware supports either
"fast" (10MB/sec) or "slow" (5MB/sec) SCSI data transfer rates. If you've turned
on Fast SCSI, make sure that the total cable length is less than about 6 feet.
This includes the cable segments inside the disk enclosures. Fast SCSI requires
shorter cables than Slow SCSI does. Make sure that the external SCSI terminator
is getting power. At least the last device on the bus must supply it. You
cannot depend on adequate termination power coming from the CPU. And check the
power supply."
Received on Thu Oct 05 1995 - 15:32:42 NZDT