MAINDEC for Alpha? - test-drb@ccmp.vtda.org

14 Jun 2011

Dave McGuire wrote:
...
 On 6/13/11 5:11 PM, Daniel Seagraves wrote:
  Is there something like MAINDEC for Alpha?
 I have an AS1000 that randomly goes catatonic and I'd like to try to narrow down
what's going on.
 It will do it shortly after boot, usually within 15 minutes, and won't run for longer
than an hour at most.
 Sometimes it hangs hard and must be power-cycled, sometimes it crashes with random
messages.
 The SRM diagnostics say nothing is wrong, even when run for very long loops. 
   I can't help you with your specific question, but as far as general 
info about the AS1000, if you find yourself spending too much time on 
it, trash it and get something like an AS4100 to replace it.  The AS1000 
is one of the crappiest, flakiest machines I've ever seen come out of 
DEC.  I've owned several, and every one of them ended up giving me problems.

I have had a fair amount of grief with my AS1000A machines but I wouldn't say
it was much more than my other DEC machines which all have their own share of
problems.

Do you ever get a real crash complete with crashdump? If so, analysis of the
dump may help pin down the problem. If not, there might still be something in
the error log.

It may be worth making sure the CPU board is seated correctly in its slot - I
have had issues with this and I assume the AS1000 and AS1000A are the same in
this respect. As far as I can tell the main difference between them is that the
AS1000A has more PCI slots instead of (E)ISA slots.

The B-Cache on the CPU board seems to be prone to failure, however it usually
shows up in the diagnostics. It can be disabled by moving a jumper and the
machine will run with reduced performance. (The memexer diagnostic will then
spectacularly fail as it appears to divide something by the cache size...)

I've also had memory failures but these also showed up in the diagnostics for
me (sometimes the diagnostics seem to report the wrong slot as being faulty).
However, I've also seen memory problems in other alphas cause random hangs
while nothing shows up in diagnostics. You could try running with a reduced
amount of memory.

As with any machine, it is worth checking the the power supply voltages
are correct and don't have excess ripple.

Right now, I've got one that used to randomly halt and now randomly switches
itself off. I suspect problems with the front panel switches or related logic
but I haven't worked up the enthusiasm to check this theory out.

Regards,
Peter Coghlan.