Syslog for High Temperature

From Cisco, you can have it ship syslog data to a system or tool such as kiwi to ingest the syslog data.  Then depending on which tool, you may be able to write rules to react to the event.  

What may be preferable and provide some reportable data is to utilize that as well as the SNMP poll for temperature, that way you can see at a glance what the temperature is and be able to graph it over time.  Just using the syslog component is useful for react to an over threshold condition.

Thanks.. My issue is that there are different models and for each model Cisco has specified different Temperature range. As  a result we thought of take the alert from the Syslog. Moreover, the SNMP is giving a lot of sub parameters which makes more confusing. If you have any good suggestion on which SNMP parameter would meet the requirement, would be helpful.

Thanks.. My issue is that there are different models and for each model Cisco has specified different Temperature range. As  a result we thought of take the alert from the Syslog. Moreover, the SNMP is giving a lot of sub parameters which makes more confusing. If you have any good suggestion on which SNMP parameter would meet the requirement, would be helpful.

and depending on the model the specific mibs are important.   What are these devices and what model?

Hello.. We have devices like C9300L, Catalyst 3560, 29xx, 2921X, 4331 ISR, Catalyst 4500, Catalyst 68xx, Catalyst 9300, NExus 7K.

Also i can see a large number of sensors for each node 

Which is suitable for the Alarm... Each giving different values.

Each sensor in the device is located in a different spot - some are closer to the exhaust, some closer to the CPU, etc. Therefore each has a different threshold. In the past I've approached this issue 2 ways:

1) SNMP polling - you have to poll the values you are finding AND ALSO the values you'll find in another table, which is the threshold for the sensor. That way you can alert when the current temperature is x% of the maximum threshold. This is complicated, and usually not as useful as...

2) SNMP traps. What you are looking for as a syslog is already present as traps coming from the device. For all intents and purposes they will operate the way you say you want, but by looking for SNMP traps you'll have an easier time finding the right one(s).

I'll also add that along with option #2, you should STILL do #1 and collect the temperature. Traps will only occur when there's a problem. You have no idea if the temperature has been slowly rising for hours, or there was a sudden spike, or something else. By collecting (but not alerting) on the temperature stat, you'll have a better picture of what is going on inside the machine.

Hope that helps.

Thanks. a lot.. Another Concern is that i could see temperature  sensors for interfaces, modules, additional cards etc.... Which is the sensor which would be ideal to select.

Moreover, can i know how to get the Threshold Values through SNMP. I have enabled Hardware Sensors and each sensor is showing current value.

Your suggestions are highly appreciated.

"which sensor would be ideal"
that depends on what you are trying to "catch". Any one of those can cause a system shutdown if they go over threshold. So from that standpoint, whichever is consistently the hottest is going to be the one you want.

On the other hand, sometimes you don't care how hot the exhaust fan is (after all, that's where all the hot air is coming out, it's going to be hot) so maybe not that one.

It all depends on what you are concerned about.

"how to get the theshold values"
The SNMP OID table "entSensorThresholdTable " (1.3.6.1.4.1.9.9.91.1.2.1) should have the thresholds for the sensors on that machine.

Using the UnDP utility on the Orion polling engine (the server itself, not the web interface) you can collect both the temperature and the threshold and then display both values in a table on the Node Details page.