rp3410 Ambient Temp Warnings

[Rod Hewitt]

I have a problem with an A9954A rp3410 PA8900 800mhz 1way hpux server. About once a week the machine sets the amber front panel light flashing. Logging in and viewing the system event log a) clears the light, and b) displays messages similar to those below:

Log Entry 469: 16 Apr 2007 11:08:52
Alert Level 3: Warning
Keyword: Type-02 010107 65799
A temperature inside the server has gone outside the factory specified range.
Logged by: Baseboard Management Controller;
Sensor: Temperature - Ambient Temp
Data1: Upper Non-critical - going high
Data2: TRG Reading: 0x25 TRG Threshold: 0x25
0x2046235944021FC0 25255701D8010300


Log Entry 468: 31 Mar 2007 07:44:41
Alert Level 2: Informational
Keyword: Type-02 018107 98567
Temperature going to normal (was warning)
Logged by: Baseboard Management Controller;
Sensor: Temperature - Ambient Temp
Data1: Upper Non-critical - going high (Deassertion)
Data2: TRG Reading: 0x21 TRG Threshold: 0x25
0x20460E1169021FB0 25215781D8010300


1 Does the value displayed (e.g. 0x25) correspond to an exact temperature? I am guessing that this means the Celsius temperature in hex, i.e. 37 degrees.

2 Where exactly is the sensor that is being triggered?

HP website suggests the following possible causes.

* A failed fan
* Blocked air vents or poor ventilation around the server
* Open server chassis cover, or missing drive slot covers
* High temperature in the room where the server resides
* An unusually hot hardware component located near one of the server's temperature sensors

Given that this machine is located quite a long way away, can I diagnose any of these remotely? I assume that a failed fan would be reported separately. I am repeatedly assured that the temperature in the room is not particularly warm. They are leaving the rack doors open to ensure that there is plenty of airflow. Other (Windows) servers in the same rack appear to be fine.

Any help or suggestions welcomed.

[goldboy]

for a start Rod you can log into the MP and run the PS command that will allow you to see the fans status.

log into the MP
MP> CM
CM> ps

if there is a bad fan it will list it in the fans and power supply status

Tal

[Bill Hassell]

I would start with a 2 dollar thermometer near the front of the server. Some people are comfortable in a room that is 35 deg C. But for computers and disks and tape drives, you can be assured that 35 is not good for your equipment, especially if they are opening the doors mto help cool the equipment. One of the worst components in computer room design seems to be the air conditioning, sometimes left to someone that guesses about the heat load and worst case heat differential. Ambient temperature is part of the measurement but also included is what else is in the cabinet and how the air in the cabinet flows out. And what happens when no one is there to fix a broken air conditioner?

Although computers are not necessarily accurate thermometers, I would not ignore this condition, especially if the outdoor temperature is higher than normal. It means that the room may be about to destroy a lot of expensive equipment -- or put it another way: prove that the room is reliable enough to protect thousands of dollars of equipment.

[Torsten.]

Hi,

these are the specs:

Operating Temperature (up to 5000 ft)*
41° to 95° F (5° to 35° C)

http://h18000.www1.hp.com/products/quickspecs/12036_na/12036_na.HTML

The stages are like this:

- critical overtemp warning
- emergency overtemp warning
- shutdown
- power off

This all happens in a small range around the max. operating temp.

Better to measure the ambient temp and adjust the air conditioning if needed.

See also
http://docs.hp.com/en/B2355-60105/envd.1M.html

[Rod Hewitt]

Thank you for the useful replies.

I have encouraged the responsible person to leave the glass door off the front of the rack, and to buy a cheap thermometer. (I had previously suggested they get one of those remote 'weather station' style clocks. I have one on my desk - cost GBP 12! Then they could check the temperature remotely without even getting up from their chairs.)

I checked power supplies/fans - all reported to be well. I had not realised that was possible.

The whole lot is going to be relocated in a few weeks so I cannot really start shouting about air conditioning, etc.

Points will be done in a while.