There are multiple ways to setup monitoring in Cumulus Linux. Nagios is what ends up in our literature quite frequently. The other alternative tool that we frequently mention is Ganglia.
If you want to track traffic flows through the box, then you will want to use sFlow:
Then you will need an sFlow collector to receive the traffic stats. Nagios is such a collector:
the key statistics that we get from the chipset are stored in data tables. These data tables are accessible through the commands we provide. For example, for interface statistics, those commands are described on this page:
That same data is also generally available via SNMP - which could be polled by shinken/Nagios.
The last part of this section gives the SNMP MIBs that we support.
As to what information is useful, the items I described in my previous email are what I would consider useful to know when managing a network: System CPU, System Mem, key interface counters, etc...
The nice thing is our boxes are Linux, and you can install 'anything'. The bad thing is you can install 'anything' :-). So it would require some testing on your part to make sure that nrpe doesn't use much CPU time on the box or cause any stability issues.
If you can collect the same information that nrpe via other methods like SNMP, then I would generally recommend SNMP. That being said, Linux/devOps tools have their advantages also.
If this is useful, here are some other tools that were recommended by an engineer that had experience with them:
- checksnmpnetint https://github.com/willixix/WL-NagiosPlugins/blob/master/check_snmp_netint.pl Very powerful to monitor IF-MIB which can warn against traffic thresholds, includes performance monitoring output too
- checkospfcounters http://exchange.nagios.org/directory/Plugins/Network-Protocols/OSPF/check_ospf_counters/details
- lmsensors http://www.thomas-krenn.com/en/wiki/LMSensorsMonitoringPlugin We present fan rpm, temperature, etc under the standard Linux lmsensors framework
- check for security updates http://library.nagios.com/library/techtips/264-tech-tip-monitoring-critical-debian-and-ubuntu-updates-with-nagios
You may also want to look into Icinga, a fork of Nagios which has more community maintenance and slightly different features while still being backwards compatible.