Temperature outside normal range - mib defines normal as?

[Sean Murray_1]

Got a few alerts from CPQSTSYS-MIB letting me know temp going back and forth from 'ok' to 'degraded' which is defined as outside of normal operating tange.
Great.
But how is it defining this? Through BIOS settings, through HP defining?

Personally I want to see a temp, not just a description. Anyway to do this?

Thanks,
Sean

[David Claypool · Accepted solution]

ProLiant servers can have 1 or more sensors that are individually calibrated. See the Environment section of the Systems Management Homepage for the specifics for the system in question.

[Sean Murray_1]

Thank you. I see it.
Out of curiosity, and not that I am going to, but are these values able to be changed anywhere?

Thanks

[David Claypool]

No, they are not changeable as they are based on the specifications of the devices they protect. However, be warned that there is some complex physics behind these data. In other words, if you were to look at the data sheet for an Intel Octium (fictitious processor for purposes of illustration) and found that it had a specification with a top operating temperature of 1000K (again fictitious: 1000 Kelvin is 1340 Fahrenheit), don't be surprised if you looked at the thresholds on the Environment page and saw that processor 1 had a warning threshold of 482K and processor 2 had a threshold of 828K. Why would that be? Because it would depend on the actual location of the temperature sensor and the amount of airflow past it at that point. In the lab they would have done measurements and performed the calculations to show what the sensors would register when the component was at a specific temperature.

If the thresholds were able to be modified by the end user, it would cause chaos as people tried to reason out what the individual settings should be--or maybe not: just set them all the same! You can see how confusing this would be and what kind of warranty situation this would cause, especially considering that some systems have 6 or 8 sensors!

The one thing you can do is to change the system's behavior on the Environment link to continue or shutdown in an overtemp condition. In either case, the agent will send an SNMP trap when this occurs. The issue is whether or not someone is on duty to act upon the alert before component failure begins.

[Sean Murray_1]

OK. Good enough and makes sense.

My only wish is that the internal logs and thge alerts would also show the actual tmep when the alarm was sent, rather than unknown.

But thanks for the helpful info.

Sean

[David Claypool]

The actual data aren't relevant (What would it mean? You can't infer ambient temperature from internal readings.). The only important piece of information is that the threshold was exceeded.

[Sean Murray_1]

Closed