[RESOLVED]Discrepancy between IMB ACPI and SMART data for HD
Posted: Sun Aug 11, 2013 7:08 pm
I am getting a huge discrepancy between the data for my HD temperature reported by IBM ACPI and the SMART data provided by the hard drive itself. The HDD temperature reported by the IBM ACPI interface, is it just "somewhere nearby the hard drive"? That would explain some difference, but it is too big!
I'm running an X61 using Ubuntu 12.04 LTS, the HDD is an Western Digital WD3200BPVT (320Gb/5400rpm/8MB Cache).
For example:
When
IBM ACPI reports 59C (cat /proc/acpi/ibm/thermal)...
SMART reports 45C (smartctl -a /dev/sda1).
So I suppose the SMART reading is the correct one? The IBM ACPI was telling me my drive was going as high as 62C, although the HDD was idling (only the CPU was working hard). The fact that the IBM ACPI HDD temperature reading is so closely correlated to the CPU temperature first led me to believe that the ventilation is so poor inside the X61 chasis around the HDD that the drive would heat from the CPU radiant heat, but running in Windows I didn't see this correlation, so then I suspected bad power management in Linux. Now, this leads me to believe that the IBm ACPI HDD reading is wildly inaccurate and simply reports the reading from a sensor "in the vicinity of the HDD", and probably very sensitive to CPU heat.
Any thoughts?
Thanks!
***** RESOLVED
The way to go is with the HD SMART data. There is a program in Linux called hddtemp which reports the HDD temperature in Celsius as provided by the SMART interface. You can run it as client (one time data output) or as daemon, so that system monitors can read realtime temperature data.
Assuming te HD whose temperature you want to check is /dev/sda1, you run it as:
- Client:
sudo hddtemp /dev/sda1
- Daemon:
sudo hddtemp -d /dev/sda1
If you start it as daemon, then run gkrellm like I do, you will have a sensors entry called hddtemp/sda1.
You need to have the hddtemp daemon running before gkrellm is started. I chose not to started via the usual startup scripts in Linux, and instead start it with the startup script for fluxbox (~/.fluxbox/startup) just before starting gkrellm.
I'm running an X61 using Ubuntu 12.04 LTS, the HDD is an Western Digital WD3200BPVT (320Gb/5400rpm/8MB Cache).
For example:
When
IBM ACPI reports 59C (cat /proc/acpi/ibm/thermal)...
SMART reports 45C (smartctl -a /dev/sda1).
So I suppose the SMART reading is the correct one? The IBM ACPI was telling me my drive was going as high as 62C, although the HDD was idling (only the CPU was working hard). The fact that the IBM ACPI HDD temperature reading is so closely correlated to the CPU temperature first led me to believe that the ventilation is so poor inside the X61 chasis around the HDD that the drive would heat from the CPU radiant heat, but running in Windows I didn't see this correlation, so then I suspected bad power management in Linux. Now, this leads me to believe that the IBm ACPI HDD reading is wildly inaccurate and simply reports the reading from a sensor "in the vicinity of the HDD", and probably very sensitive to CPU heat.
Any thoughts?
Thanks!
***** RESOLVED
The way to go is with the HD SMART data. There is a program in Linux called hddtemp which reports the HDD temperature in Celsius as provided by the SMART interface. You can run it as client (one time data output) or as daemon, so that system monitors can read realtime temperature data.
Assuming te HD whose temperature you want to check is /dev/sda1, you run it as:
- Client:
sudo hddtemp /dev/sda1
- Daemon:
sudo hddtemp -d /dev/sda1
If you start it as daemon, then run gkrellm like I do, you will have a sensors entry called hddtemp/sda1.
You need to have the hddtemp daemon running before gkrellm is started. I chose not to started via the usual startup scripts in Linux, and instead start it with the startup script for fluxbox (~/.fluxbox/startup) just before starting gkrellm.