[comp.sys.sun] Power failure wedges SCSI disks on SparcStation I's

MIKE@ucdphy.ucdavis.edu (Mike Hannon; UCD Physics; (916)-752-4966) (10/08/90)

We have a small network of Sun systems, including a Sun 4/280, 2 Sun
4/110's, and 3 Sun 4/60's.  These are all in the same building as, and
connected to the same ethernet as, a dozen or so VAX machines.  All the
Suns are running 4.0.3 (or 4.0.3c).

Recently we had a power glitch (I was not here at the time it happened, so
I can't describe it any more fully).  This caused all the computers to
crash.  The Sun 4/280 and one of the 4/110's rebooted spontaneously, as
did most of the VAXes.  The other 4/110 had some difficulty with the NFS
initially, but finally booted properly, without any action on our part.

All three of the 4/60's failed to boot.  Each of them is configured with two
internal, 104(?)MB drives, installed by Sun.  In all three cases we got the
following messages on the console:

			Booting from sd(0,0,0)vmunix
			Can't read disk label.
			Can't read boot blocks.
			    Boot load failed.

I attached a shoebox tape drive to one of the 4/60's and booted memory
Unix from the 4.0.3c distribution, expecting to have to format sd0.  When
I ran format, it found only sd1, not sd0.  At that point I was convinced
that we had a hardware problem, and I summoned help from the campus
hardware gurus.

A guru showed up, removed sd0, and took it back to his lab to check it.
When he did so, the disk ran perfectly, original contents completely
intact.  His diagnosis was that the disk was simply not spinning when it
was in our machine, but taking the disk for a walk made it work!

Sure enough, after he removed the disks from the remaining 4/60's, shook
them gently, and replaced them, the systems booted flawlessly.

Thus, it appears that some kind of a power surge made all three of the Sun
SCSI disks get ``wedged'' (literally).  A gentle shake restored them all
to proper working order.  The second SCSI drive on each machine seems to
have been unaffected by all this.

Just for my information: has anybody else seen this?  Does anybody have
any idea as to what is happening here?  Thanks.

Mike Hannon                              mike@ucdhep     (Bitnet)
ucdhep::mike          (HEPnet)           42385::mike     (HEPnet)
jmhannon@ucdavis.edu  (Internet)         916-752-4966    (Telephone)