We ran into a series of shutdowns today with our Sun W2100Z dual
Opteron running Fedora Core 4 and the 2.6.12-1.1456_FC4smp kernel.
The system log showed...
Oct 24 13:22:44 XXXXX kernel: Critical temperature reached (68 C), shutting down
.
Oct 24 13:22:44 XXXXX kernel: Critical temperature reached (68 C), shutting down
.
Oddly, when these shutdowns happened both cpu fans were running normally.
Unlike the surged fans that occur when the machine boots from the bios
we saw no evidence of the fans surging suggesting there was no temp problem.
We finally upgraded to the latest 2.6.13-1.1532_FC4smp kernel and so
far these shutdowns have ceased.
Does anyone know if there was any recent bugs in the acpi code of the
kernel which could have been causing spurious shutdowns due to critical
temperatures? Looking at this machine under full load on both cpus, I
see...
cat /proc/acpi/thermal_zone/THRS/trip_points
critical (S5): 65 C
and
cat /proc/acpi/thermal_zone/THRS/temperature
temperature: 44 C
Unfortunately, lm_sensors is non-functional on this motherboard because
the driver for the ADT7467 chip that controls the fans and temperature
doesn't exist yet. So I can only go by the observed fan speed and the
temperature as reported by /proc/acpi/thermal_zone/THRS/temperature.
However I am unclear if this is a cpu or a motherboard temperature. If
it is a cpu temperature it is unclear if it is of just one or an average
of both.
Thanks in advance for any advice on these issues.
Jack
