[rescue] Catatonia in the SGI Indigo2
Sheldon T. Hall
shel at cmhcsys.com
Fri Jan 16 10:31:05 CST 2004
David L Kindred (Dave) writes
> >>>>> "Sheldon" == Sheldon T Hall <shel at cmhcsys.com> writes:
> Sheldon> It hasn't done this while running Distributed Folding hour
> Sheldon> after hour, ...
> I'm suspecting that is lots of CPU but little disk activity?
Correct-oh. It updates its status and log files occasionally, but that's
only a few bytes every few minutes.
The FTP and SMB file transfers that failed involve lots of disk activity,
but burning a CD (on an external drive) from WAV files (on an internal
drive) doesn't involve all that much.
> Sheldon> ... but it did it several times while accepting SMB file
> Sheldon> transfers and FTP file transfers. ...
> Sheldon> ... Just now, it crapped out while I was watching it burn
> Sheldon> a CD from some WAV files on the internal drive.
> Based on this, I'd suspect the power supply. It's hanging on the edge
> most of the time, but when the load peaks it just can't handle it.
Do you know of any way I can test that? I can put a voltmeter on the
various power leads, I suppose, though I'd be hard-pressed to see a sudden
and transitory dip in the voltage that way.
> Of course if the dust issue brought by others is the case, then the
> additional thermal load of the increased drive activity could be at
> fault. Do these machines have a thermal cut-out mechanism? Double
> check that all of the fans are running all of the time.
My Challenge L has a neat-o "system controller" that controls the fan speed
(and the fan is _huge_), software that will report the system's temperature,
and a thermal cut-out that will gracefull shut the system down. AFAIK,
however, the I^2 has none of the above. Certainly, it's not shutting down,
gracefully or otherwise. It's just hanging. I suppose the front-panel
pilot light's changing from green to yellow means something, but I haven't
been able to establish what.
More information about the rescue