Nagios performance results...


20-01-2005 14:24:17

We originally looked into NetMRG as a replacement for Nagios..
As it seems it isn't a full blown solution to notification yet.. (We perform pings in 1 minute intervals) and there isn't a *DIRTY WORD* windows client to monitor specific things like if a certain process is running...
(which might be possible with windows snmp or ucd-snmp.. I havn't investigated it..)

Has anyone looked at writing a test script to parse to parse the Nagios performance data logs?


20-01-2005 15:05:56

Right; NetMRG isn't quite a Nagios replacement yet, but we are working in that direction.
I'm not aware of any scripts right now that parse nagios perfdata; I don't think it should be that difficult to write, though. The biggest problem would be automating the creation of services in both tools so perfdata can be graphed easily.



20-01-2005 15:16:53

I'm pretty crafty with SQL.. I think I'll take a look at the schema and see how hard an automation process will be..


18-02-2005 14:20:24

I stumbled upon this thread in a web search, and will need to look at NetMRG
at some point, but I have a fairly satisfactory Nagios setup. In addition to the
standard pages/alerts when something fails, I also collect a fair amount of
quantitative data on selected services using the perfdata option.

I periodically (about every 15 minutes or so) have a cron job read the data and
stuff it into round robin databases (using the Perl interface to rrdtool). Add
another home-grown script to generate graphs from the rrd files (using rrdtool
package, but I did not like rrdcgi tool) and some static web pages, I have a nice
historical trends graphs a la the old MRTG tool.

The configuration files for these tools are somewhat lengthy and repetitiious,
e.g. the tool to read the Nagios perfdata logs wants to know what variables to
read for each host/service combo, and all the information needed to create the
rrd file to dump the information into. Similarly the cgi code for graphing wants
for every possible graph to produce all the arguments to be given to the
rrdgraph command. Rather than attempt to change/simplify the input to the
tools, which would likely reduce flexibility, I have set up Makefiles and use the
m4 preprocessor to eliminate much of the repetition, so I perfect my configuration for a certain type of server/service, then use m4/make to generate
the configs for each host/service needing it.

If you still are looking for this sort of stuff, let me know and can show you what
I have.

Tom Payerle