lm-sensor and hp-ux

[rafalcaide]

Hello, i try to install lm-sensor in a hp-ux system, to monitor temperature, without success.

¿How monitoring temperature in hpux?

Thanks.

[Torsten.]

How about a thermometer?

;-)

Most server models can only give you status like OK, overtemp or critical overtemp - no exact values, but it depends ...

[Pete Randall]

Some UPS's (our APC, for example) have the capability to monitor temperature.


Pete

[Torsten.]

HP-UX is monitoring the temp, see

http://docs.hp.com/en/B3921-60631/envd.1M.html

You can try to get the status from the GSP/MP/iLO - some examples:

BL860c:
MP:CM> ps


PS
For System Processor Status see the SS command.

For Power Supply and Fan status see the Onboard Administrator.

System Power state : On
System Power usage : 203 Watts
System Power allocation : 255 Watts
Ambient temperature : 24 C
Temperature status : Normal

Over Temperature protection: Enabled


rx2660:

MP:CM> ps


PS
For System Processor Status see the SS command.
System Power state: On
System Power usage: 239 Watts
Temperature : Normal

rp7420:

MP:CM> de

Display status of the selected MP entity (for use by trained personnel only)

B - BPS (Bulk Power Supplies)
U - CLU (Cabinet Utilities: Fans, Intrusion, Clock's etc.)
A - PACI (Partition Console Interface)
G - MP (Management Processor)
P - PM (Power Management)
H - Cell Board Controller (PDHC)
Select device: u

Cabinet 0 Utilities Status
FW Revision : 8.005 built Sep 26 2006 at 16:11:20

PWR SBY MP RUN REM ATT FLT
Front Panel LED State : * * * . . . .

Inlet Air Temperature : 20 deg C
Altimeter Status : OK
Operating Altitude : 2000 - 4000 ft (608m - 1216m)

Fan Rate ( rpm/30 ) : Standby Main I/O
97 97 95
...

[rafalcaide]

Ok, but i try monitoring hpux temperature from nagios.

[Duncan Edmonstone]

Why do you want to monitor the internal temperature of a server? It's no indication of your machine rooms ambient temperature, and HPUX will take action anyway if it goes over-temp

HTH

Duncan

[Torsten.]

You may try to configure the ems to send snmp traps for overtemp events, but finally you will see the server shutdown and power off, if temp is too high.

[Bill Hassell]

As mentioned before, using computers to measure temperature is a very bad idea. They are lousy thermometers. Older boxes have no temp sensors, others just have a high-temp and over-temp which is not calibrated and the actual temperature varies between models.

There is only one reason to monitor temperature and that is to provide alerts in case the data center overheats. Most newer HP-UX computers have variable speed fans and will try to keep themselves cool in case there is a fan failure. If the ambient room temperature is too high, the HP-UX boxes will probably survive more than 100 Deg F all your disks, tape drives, printers, LAN switches, routers, firewalls, etc will probably be fried.

The other problem is that Nagios only works in a *LOT* of equipment is working properly. The Nagios server must be up 24x7, the network switches must work, the email realy system must work for sending alerts and finally, the email recipients must always monitor their messages 24x7. That is an incredibly risky design. One failure (most commonly the nut at the end of the email chain) to read and act on the overtemp means thousands to millions or dollars in damage.

Is your building monitored for fire and water damage as well as for security? Do what most small data center managers do: purchase temp sensors designed for a monitoring service and tie them in. Now you have a low-tech telephone (but highly reliable) chain to someone that will keep calling people until someone answers and agrees to takes action. None of this is possible with a one-way notification system.

I know that Nagios and computer sensors sounds cool but I have witnessed too many data center disasters caused by broken air conditioning systems that failed to get any attention. For the data center, temperature, water and fire are all the same importance and must be handled in the same way.

[rafalcaide]

I should monitor the temperature of the cpu a rp 7000.

Is it possible from nagios?

[Torsten.]

I think it was clear enough already - you can only monitor (even if) a status like "normal/warning/critical/down/off" - but no degree values.

[rafalcaide]

Thanks Bill Hasell, we will not to disable over_temp method, but we had installed a monitoring system based on nagios, and we want monitor all system and services as posible.

We can be able to monitor linux server temperature with lm-sensor but we had not found this package for HP-UX systems.

Thanks to all for reply.
Sorry for my English.

I will continue investigating...

[Torsten.]

I don't know much about lm-sensor, but I read that it is made for linux, pc style hardware (probably not RISC or Itanium) and works close to the linux kernel.

As suggested above, how about sending traps? IMHO this should be possible.