[Info-vax] Another Alpha "TEMPERATURE_VECTOR" anomaly

Rich Jordan jordan at ccs4vms.com
Fri Aug 24 11:04:22 EDT 2012


On Aug 23, 10:37 pm, moro... at world.std.spaamtrap.com (Michael Moroney)
wrote:
> Rich Jordan <jor... at ccs4vms.com> writes:
> >Fan test shows the wonky sensor does respond to ambient changes, and
> >more or less tracks with the working one.  A fan aimed at the front
> >panels dropped the logged temperature on both units by the same amount
> >(4 - 5C).  If the warmup test does the same then we'll just set the
> >thresholds lower on that system's checks, and if needed add a power-
> >off shutdown at XX degrees C (to be determined but around 22C lower
> >than the working system).  Easy peasy.  We've had to do a hell of a
> >lot more to keep various wintels running reliably over the years.
>
> Looks like you found what I was about to tell you.  If the sensor works
> but consistently reads 22C low, you can set (from the console) the
> shutdown temperature to be 22C lower.  I don't remember how to do it, but
> I did do that before on a DS10L. (pushed my luck on a system that would
> overheat on hot summer days in an unairconditioned room by raising the
> temp 2C.  I risked frying the CPU but hey, the system was free from
> Islandco and was a hobbyist system so I wouldn't be out anything.  The CPU
> survived my abuse)

The DS10L RMC shows warnings at 55C and power-off at 60C; I assume
those are defaults since that matches examples in the console
reference guide.  But I don't see any way to change them in RMC or the
console; there is a 'shutdown_temp' environment variable in the
console set to 60C but it appears to be read-only.

There was a thread on ITRC which resisted earlier searches:
http://h30499.www3.hp.com/t5/Hardware/Alphaserver-DS10-auto-shutdown/td-p/3813712#.UDeWEKCGyVc
where a user had the opposite problem; his reported temps 20+ degrees
warmer than actual.  There was one mention at the end about creating a
console environment variable called 'over_temp' to modify the
settings, but no followup.

Regardless I set that on the test DS10L setting it to 57C to see what
happens.  RMC does not show any difference but it may require a full
restart or power cycle to take effect; we'll find out at lunchtime.



More information about the Info-vax mailing list