ACPI / cpu temperature problem
Hi, recently we bought 12 identical Supermicro servers. On these we run Fedora Core 2 with all available updates through yum.
On most of our other servers we have the ability to monitor the cpu temperature through '/proc/acpi/thermal_zone/THRM/temperature' and this works fine. We monitor most servers using rrd graphs.
Now, the problem with all these supermicro servers is that the temperature seems to be stuck at 27 C. No matter what load or temperature in the room. Something is clearly wrong. What can be done to fix this? We tried setting polling_frequency to '10', but that made no difference.
What can be done to fix this? Btw, lm_sensors finds no chips.
The specs of the model can be found here:
http://www.supermicro.com/products/system/1U/5013/SYS-5013C-T.cfm
Some maybe useful info follows:
-------------------------------
# cat /proc/version Linux version 2.6.5-1.358smp ([email protected]) (gcc version 3.3.3 20040412 (Red Hat Linux 3.3.3-7)) #1 SMP Sat May 8 09:25:36 EDT 2004
# cat /proc/acpi/thermal_zone/THRM/temperature temperature: 27 C
# lspci 00:00.0 Host bridge: Intel Corp. 82875P Memory Controller Hub (rev 02) 00:1e.0 PCI bridge: Intel Corp. 82801BA/CA/DB/EB/ER Hub interface to PCI Bridge (rev c2) 00:1f.0 ISA bridge: Intel Corp. 82801EB/ER (ICH5/ICH5R) LPC Bridge (rev 02) 00:1f.2 IDE interface: Intel Corp. 82801EB (ICH5) Serial ATA 150 Storage Controller (rev 02) 00:1f.3 SMBus: Intel Corp. 82801EB/ER (ICH5/ICH5R) SMBus Controller (rev 02) 01:09.0 VGA compatible controller: ATI Technologies Inc Rage XL (rev 27) 01:0a.0 Ethernet controller: Intel Corp. 82541EI Gigabit Ethernet Controller (Copper) 01:0b.0 Ethernet controller: Intel Corp. 82541EI Gigabit Ethernet Controller (Copper)
# sensors No sensors found! # sensors -v sensors version 2.8.6
during boot:
ACPI: Fan [FAN] (on)
ACPI: Processor [CPU0] (supports C1)
ACPI: Processor [CPU1] (supports C1)
ACPI: Thermal Zone [THRM] (27 C)
ACPI: (supports S0 S1 S4 S5)
ACPI: Power Button (FF) [PWRF]
Sincerly
Hans K. Rosbach
Hi,
Quoted 7 lines · Hans Kristian Rosbach
Now, the problem with all these supermicro servers is that the temperature seems to be stuck at 27 C. No matter what load or temperature in the room. Something is clearly wrong. What can be done to fix this? We tried setting
polling_frequencyto '10', but that made no difference.
To confirm this I've found this:
Last kernel I used was 2.6.7-rc1-mm1. There "acpi -t" was reporting correct temperatures (as lm-sensors). Now with 2.6.7-rc2-mm2 It reports 22C constantly. lm-sensors gives 40-43C.
This is on nForce2 MB with Athlon XP2500+
Same .config.
Lenar
Quoted 5 lines
Now, the problem with all these supermicro servers is that the temperature seems to be stuck at 27 C. No matter what load or temperature in the room. Something is clearly wrong. What can be done to fix this? We tried setting
polling_frequencyto '10', but that made no difference.
I reported this to the kernel bug tracker, but there seems to be no forward movement at all. So I'll try here again in the hope that someone that know this code atleast has a comment to it.
http://bugme.osdl.org/show_bug.cgi?id=2855
Sincerly
Hans K. Rosbach
Quoted 8 lines · Hans Kristian Rosbach
Quoted 7 lines
Now, the problem with all these supermicro servers is that the temperature seems to be stuck at 27 C. No matter what load or temperature in the room. Something is clearly wrong. What can be done to fix this? We tried setting
polling_frequencyto '10', but that made no difference.
Quoted 7 lines
I reported this to the kernel bug tracker, but there seems to be no forward movement at all. So I'll try here again in the hope that someone that know this code atleast has a comment to it.
Forwarded some time ago this and my own letter to [email protected] too, but no reaction there either ...
Lenar
Quoted 11 lines
Quoted 5 lines
Now, the problem with all these supermicro servers is that the temperature seems to be stuck at 27 C. No matter what load or temperature in the room. Something is clearly wrong. What can be done to fix this? We tried setting
polling_frequencyto '10', but that made no difference.I reported this to the kernel bug tracker, but there seems to be no forward movement at all. So I'll try here again in the hope that someone that know this code atleast has a comment to it.
With some help from the very nice and efficent (!!) guys at Supermicro, the problem has been resolved in bios for now by removing the cpu temperature feature. This atleast solves the problem for us, since it now no longer advertises a fake temperature.
The board P4SCE/P4SCA bios version 1.2c (or higher) will soon be released
officially with this bugfix.
Hopefully they will include support for actual temperature readout at some time.
Supermicro staff: Thank you =)
Sincerly
Hans K. Rosbach