Will a temperature-related shutdown leave a log trace?

[Mike Carlson_3]

Three servers have been shutting themselves off seemingly at random. There are no log entries in the Windows Event Logs or IML to indicate why, but the Windows logs indicate that they are graceful shutdowns, not crashes.

I suspect that this is a temperature issue, as these three servers consistently run 10 or more degrees C warmer than other servers in the room, although they are usually at least 10 degrees below the thresholds as well.

Do I just need to improve cooling/airflow in the server area, or does the lack of log entries indicate this may be something else?

The servers in question all run Windows 2000 and Citrix Metaframe. Two of the servers are DL360s, each with dual 3.2GHz processors and 2GB of RAM. The third one is an ML370 with dual 1GHz and 1.5GB.

[Bostjan Kosi]

You could just be right. As much as I can remember, there is no log in IML nor is there an ASR event reported when treshold temperature is reached. If this boxes are loaded with work, just provide them with enough cold air on the front.... I do not think that 3 of the same kind would have the same HW defect...

BR
B.

[Alasdair Clark_1]

I have experienced thermal shutdowns and can confirm events are logged to the system log by the insight agents, below are a couple of examples :-

source: cpqasm2 Event ID:4111 type: Error

Description: "Environment Abnormality Auto Shutdown (EAAS) initiated due to thermal reasons, either resulting from the system overheating, or from the loss of cooling. "

source: cpqasm2 Event ID:4120 type: Warning

Description: "The temperature (thermal sensor #3) of the system has exceeded the threshold. "

source: Server Agents Event ID:1135 type: Warning

Description: "System Information Agent: Health: A Temperature Sensor Condition has been set to degraded. The system may or may not shutdown depending on the state of the thermal degraded action value '3'.
Chassis: '0'; Location: '6'
(Thermal degraded action values: 1=other, 2=continue, 3=shutdown)
(Location values: 1=other, 2=unknown, 3=system, 4=system board, 5=I/O board, 6=CPU, 7=memory, 8=storage, 9=removable media, 10=power supply, 11=ambient, 12=chassis, 13=bridge card)
[SNMP TRAP: 6041 in CPQHLTH.MIB] "

[Mike Strako]

Agree. You will get alerts IF you have the insight agents installed and running on the servers and updated (very important). Information described above are comming from the agents MIBS.

[Paul Lammertyn]

I'm currently installing a 'new' ML350 G4 with Windows SBS 2003 and it has a faulty heatsink or thermal detection of some kind. It shuts down every 30 minutes or so and ... it DOES leave several traces, at least in my case.

I have event ID '4120' from source 'cmpasm2' in the windows systemlog with description: 'the temperature (thermal #2) of the system has exceeded the threshold.

followed by event ID 4111 saying 'Environment Abnormalmality Auto Shudown (EEAS)initiated due to thermal reasons...

I can also see events logged in the HP Proliant Integrated Management Log Viewer. 'Automatic Operating system shutdown initiated due to overheating condition.

[Michael Saurman]

Sometimes faulty memory can cause a server to bounce without any log entries made.

[Claas Kruhoefer]

there should be also entries in the integrated management log file on the server. At least at one of my servers which reportet heat issues, it has sent trap via SNMP and logt it in the log also.

[Mike Carlson_3]

I updated the Management Agents on the servers to the latest version, as well as cooling the server room a litte better.
The servers have remained up for the last three days without problems.