Over Temperature Issue | Please help

[NKHARB]

Hi All,
We have a situation here where our HP-UX 11.0 Box becomes non responsive and then the only way to rectify the issue is to reboot the server. It happened 2 days in succession and it never used to happen earlier. When i checked the syslog, i found the entry " _CRIT threshold. Correct the over-temperature condition immediately.". I found this entry for both the days when our server became non responsive. Can you please let me know how to correct this issue on our server. I new to Unix world and would appreciate if you can help me out.
Thanks,
Naveen

[Mridul Shrivastava · Accepted solution]

Could you please check the file /var/opt/resmon/log/event.log, Is there any event about the overtemp..

I suspect there is some envirnmental issues which is causing temperature increase and server is panicing.. Need to check room temperature and rectify the cause.

[Patrick Wallek]

Overtemp means that the air going into the machine is TOO HOT! "_CRIT" means that it is CRITICAL and the machine is about to be shut down.

The way to correct this is to make the room housing the server cooler.

What type of server do you have? Is the room with the server very hot? If so, that is your problem. If you have a thermometer in the room, check the temperature when you notice this problem.

[A. Clay Stephenson]

Well, I am going to take a wild guess and say that you need to correct the over-temperature condition. There is no adjustment for these values. Obviously, the first thing that you need to do is determine the ambient temperature. Is that value within the acceptable range? If the room temperature is ok then you need to check for blocked/dirty vents and filters and under-performing fans. You don't bother to mention the type of hardware you are running so it's difficult to be very specific. In any event, this is not a UNIX problem but rather a thermodynamics problem.

[Bill Hassell]

And just to let you know, an over-temperature condition will destroy your computer. There does not have to be a fire burning the computer, just let it sit in a broken computer room where the temperature is over 50 degrees C (120 deg F) and your computer will need very expensive repairs if not a complete replacement. Now certain newer models of HP 9000 boxes (not all) have the ability ot power themselves off when the temperature is too high but your disk, tapes, routers, printers, etc will be damaged (and possibly already damaged). Air conditioning in a computer room is not a luxury, it is a mandatory requirement and if uncontrolled, even for a few hours, you may faced with a total loss of all your equipment.

[A. Clay Stephenson]

... and the rest of the story is that you may have already done irrepariable damage. A common exercise in Intermediate Electronics Lab is to measure the characteristics of a semiconductor device (typically a junction transistor) and then intentionally subject the device to an brief, one-time over-temperature condition. The device is now allowed to cool back to the original temperature and the same measurements are taken. You will find that the device's characteristics have been permanently changed.
Often, components that have been allowed to overheat will work almost perfectly --- absolutely the worst kind of problem to troubleshoot.

[NKHARB]

Thank you all for your replies. I'll look into it ASAP and will do the needful. Once again, thanks for your help. And yes, i have awarded points to all of you :)

[NKHARB]

I'll check the server room and will check the temperature maintained as well.