[mrtg] Re: latency reports

[Leif Neland]

> I am currently running some sample latency reports on MRTG with the method
> described at http://www.lns.com/papers/mrtg/.  Right now, I'm only doing
> this for 5 nodes.  Each one gets 5 pings sent to it in 1 second intervals,
> so it takes 25 seconds to finish.  But if I add, say, hundreds of nodes
> and send 5 pings to each, it would take forever to finish.  If I have a
> cron job set to run this every 5 minutes, a new MRTG will be fired off
> before the old one has finished, and it's not possible to run more than
> one MRTG process at the same time.
>
> That being said, is there a more efficient way of sending out pings for a
> latency report via MRTG (aside from setting up multiple MRTG servers to
> handle the load)?
>

Don't run the script from mrtg, create a separate script for each couple of
nodes, and have the script write the results to a file. Then let mrtg read
this file to get the results.

And/or use fping in a script, which calls fping 5 times and averages the
result. An exercise in perl/awk...

Leif