I was having what seemed like some SCSI problems on a DEC 3000/300LX
running NetBSD 1.6-current, so I decided to move the disks over to a
DEC 3000/500S to see if the problem lay with a problematic disk or
with the system.

After setting up and booting the DEC 3000/500S, I had to leave it to
attend to something else.  When I checked a couple of hours later, I
noticed about ten of the following error in my syslog:

    Warning: received processor correctable error.

A cursory glance doesn't reveal any particular clustering of these
errors.  They don't occur all at once, or at a particular periodic
time.  And, with ten in 8+ hours of uptime since boot, they're not
exactly coming fast 'n' furious. :-)

What exactly does this error mean?  Is it reporting an ECC memory
correction?  This DEC 3000/500 used to run Digital Unix and, regular
as clockwork, every time the machine booted it would report one ECC
memory correction as part of the boot up phase (or shortly
thereafter), but never more than that.  There were also more details
about the correction, too, under Digital Unix.

I guess I should run the PROM tests to try and see if any of the
memory is actually bad.  Is it possible to find out from NetBSD which
memory module generated the error?  I think there is a DEC 5000/600
around somewhere that I might be able to swap some memory with.

I'd like to migrate this system over from the DEC 3000/300 to the
3000/500 (dual 25 MHz SCSI buses and meatier I/O bandwidth, for one!),
but not if it seems like the hardware is going to up sticks on me.

If anyone can help explain the processor correctable error warning in
greater detail, I'd much appreciate it.

Cheers,

Paul.

e-mail: paul@gromit.dlib.vt.edu

"Without music to decorate it, time is just a bunch of boring production
 deadlines or dates by which bills must be paid."
        --- Frank Vincent Zappa