We would like to monitor the temperature on our Dell servers. I've used the built-in OIDDLIBs as well as a smaller one supplied in this forum. When I apply these to a live server I get anywhere from 2 to 6 different sensors. All of them are labeled like:
cooling device: 1/cooling device reading: 1
Which means nothing. How do I tell what each of the sensors is actually measuring? It is CPU2? The BMC Ambient temp? Planar or riser temp?
Also, from reading the MIB file I know that each of these temp sensors has settings for high and low warning and critical settings. Does the OIDLIB automatically used these settings to set thresholds for the sensor?
If not I'd like to formally request a Dell temp sensor that applies the proper string name to the sensor and uses the hi/lo thresholds from the mib for that sensor.
Something that's in the works perhaps?
Article Comments
Hi Jim,
I found these OID's and testing them on my Dell Poweredge, they seem to be producing relevant data.
| Name | OID | My value** |
|---|---|---|
| Dell Poweredge CPU1 Temperature | .1.3.6.1.4.1.674.10892.1.700.20.1.6.1.1 | 33 |
| Dell Poweredge CPU2 Temperature | .1.3.6.1.4.1.674.10892.1.700.20.1.6.1.2 | 33 |
| Dell Poweredge Ambient Temperature | .1.3.6.1.4.1.674.10892.1.700.20.1.6.1.3 | 18 |
| Dell Poweredge Planar Temperature | .1.3.6.1.4.1.674.10892.1.700.20.1.6.1.4 | 30 |
| Dell Poweredge Riser Temperature | .1.3.6.1.4.1.674.10892.1.700.20.1.6.1.5 | 24 |
Our servers are kept rather cool.
Return value is in degrees Celcius and has to be divided by 10.
Note: Dell Openmanage has to be installed on the server.
Regards,
Jan, 2012 - Permalink
Yes, I am aware of the proper OIDs. However, depending on the specific server model and the version of OpenManage that is installed these results can vary widely. For instance, when you walk the table you will get different results for servers with a different number of CPUs or Cores. This will change again if hyperthreading is enabled. Also, some Dell servers do not have the Planar, Riser or Ambient probes meaning we will not get those results.
And worse, when you use the PRTG supplied Dell SNMP libs, including the short one supplied on this forum, you only get a list of temp sensors (of variable number) with no idea which value belongs to CPU1 or CPU2 (or 3 or 4), or any of the riser/planar/ambient as listed above. Additionally, on advanced servers there are probes for the RAID arrays and others. So if I only wanted to track CPU and RAID temps and the OID library returns 8 temp values, how do I know which ones to keep and which ones to ignore? The sensor labels are available if I manually crawl the Dell table and then match up the index values one value at a time. A two week job at least for an intern. What I'd like is for PRTG to crawl both the temp value tables and the temp sensor label tables and match them up similar to what is done with SNMP interface or CPU utilization sensors. Additionally I'd like that all found temps be plotted on the same graph (just like WMI CPU and Disk graphs) and finally a check box or formula we can apply to convert temps from one scale to another without having to create an additional factory sensor.
What I'm asking for is that the supplied libraries use
Jan, 2012 - Permalink
One more request along these lines. The OpenMange Temp table als includes warning and critical temp thresholds for upper and lower temp values. In addition to grabbing the temp sensor label it would be nice if the Dell OIDLIB could also set the upper & lower warning/cricital levels based on these table values.
Jan, 2012 - Permalink
Preferably with a checkbox that allows conversion from celsius to Fahrenheit.
Jan, 2012 - Permalink