Environment (Temp) Monitoring

[Walter Keating]

Hi !

Does anyone know of a way to access current environment temperature information that EMS has available using scripts etc running on the system (HP-UX 11.0 N, L and A classes) ?

I want to check the temperature frequently and
react to AC problems before the EMS logs threshold breaches the system shutdown !

thanks

[Ravi_8]

Hi,

I ahven't seen this in HP, whereas in Tru64 system displays the current temp at boot time.

we had the similar problem(machines shutting down due to temp), adding the additional AC is the solution.

[Luc Bussieres_1]

We had a similar problem last week and looking at the log EMS will report 2 level of errors the first one (critical) will give you some time to react. and the second one (emergency) will shutdown the system.

Jul 2 15:40:32 jupiter /usr/sbin/envd[1284]: ***** OVERTEMP_CRIT WARNING *****
Jul 2 15:40:32 jupiter /usr/sbin/envd[1284]: Temperature exceeded the normal operation threshold.
Correct the over-temperature condition.
Jul 2 17:41:24 jupiter /usr/sbin/envd[1284]: ***** OVERTEMP_EMERG WARNING *****
Jul 2 17:41:24 jupiter /usr/sbin/envd[1284]: Temperature exceeded the OVERTEMP_CRIT threshold. Cor
rect the over-temperature condition immediately.

Luc,

[A. Clay Stephenson]

At best on SOME HP boxes you can get a status but not what amounts to an analog readings.

Your better bet is to purchase a digital thermometer with an RS-232 serial interface. I use a model made by Ex-Tech that can monitor two temperature inputs. A small bit of C, Perl, or even shell script to read from a serial port and you are done and you don't have to worry about finding out how to do it on a particular computer model.

My version logs alerts into VP/O.

[RAC_1]

I remember using some command to monitor fan power sully, temp etc in V class server.

But that is not possible on L, N class.

The only way is to monitor critical, emergency EMS warnings in syslog file.

You can configure EMS to do the shutdown/reboot
at different levels of warnings.

Regards,

[MANOJ SRIVASTAVA]

Hi Walter


You amy also look in this thread :

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x702feea29889d611abdb0090277a778c,00.html

Manoj Srivastava

[Bill Hassell]

With multiple systems, you probably have a fairly large computer room. Computers make lousy temperature sensors! And unless there is someone monitoring the notification 24x7, the room can still overheat and cause a disaster for all the equipment that cannot turn itself off. Even a pager may be too late if there is a delay in getting to the computer room. Some rooms are so overloaded that if the cooling stops, the room can rise to 150 deg(F) is just a few minutes. I've personally had to scrap over $100k in equipment that was damaged due to failed air conditioning.

The N, L and A calss servers are smart enough to turn off when the temperature is way too high but all your disks and tape drives and printers and modems and network routers, etc, can be severely damaged. Even though the N, L, and A class machines can turn off before they meltdown, some damage is possible and will show up as flaky operation, unexplained hardware failures and memory errors.

Add the cost to replace all the equipment and then see if 1% is a reasonable insurance policy for all this expensive equipment. Then get a good air conditioning system designed explicitly for computer rooms. These systems have multiple stages to monitor (and report temperatures electronically) as well as switch to backup systems.

Or at the very least, buy a thermostatically controlled circuit breaker for the entire room. If the temperature rises above 100 deg, kill all the power. Cheap, not nice to the applications, but certainly better than a meltdown.