Questions about status counters

[Tamás Becz]

Hi,

I'm trying to collect some statistics about my freeradius servers with nagios. Before I've been doing this with some perl code digging through the logs, and doing stats (plus generating gnuplot graphs out etc) but I'd rather have something more flexible, so I tought I'd put together some small script for nagios that can query those stats more flexible. Basically I just want to see successful and unsuccesful logins, maybe home server deads and alives, nothing really fancy. We are actually just proxying requests so I just need to see the ammount of logins through us, and the "health" of the service (we have had trouble with the party doing the real authentication, and high faliure rate is a good sign I can look for).

I can set up the status server all right, but I'm a bit stuck with how to interpret the things I get. I've been through some searching of docs, wiki and mans, but couldn't get a pointer.

1) Is there some documentation on the values I got there? I of course see dictionary.freeradius, and the names are pretty self-explanatory, but we all know devil is in the details :)

2) If I understand well, then these counters are simply monotonously increasing values. I can live with that of course, and do the math. I just like to see if my assumption is correct.

3) Also, I see that if the server is restarted, the values are reset to 0. Can freeradius be told not to do this, but retain the values? Currently we do restarts on a regular basis now and then, to update a clients.conf file. I suppose I more or less could handle this, but at least would lose the stats that happened after the last check, but before the restart.

4) I'm I looking at the good direction at all, or should I rather start reading up how accounting works or something else?

Really, I'm good with and RTFM as long as you could point me to where to look :)

(Oh, I'm using 2.1.10, but plan to update to 2.1.11 in the near future).

Thanks!

tamas