>
> Searched the archive and didn't see this addressed....
>
> We just upgraded out radius server from 1.16 to 2.0 today, and it seems to
> be working pretty well (the server is a Linux box running the 2.0.30
> kernel). There's only one problem:
>
> The radius daemon seems to be dying every couple of hours. I've set up a
> monitoring process to detect when it dies and restart it, but this is far
> from an optimal solution. As the server supports authentication for a
> couple of thousand people, even a minute of down-time will be frustrating
> for a lest a few of them.
>
> I'm running 2.0, complied from source. The compilation was clean, using
> GCC 2.7.2.1 (linked with libc 5.3.12 and libdb 2.0).
>
> Anyone else seen this behavior? And if so, any work-arounds? I've seen
> mention of "unofficial" radii in the archive.
>
> Thanks in advance!
>
> --Mark
>
>
>
YEP! Bummer! It blew the shit out of my system too! My system worked
fine for many months until I upgraded to Linux 2.0.30 (from 2.0.29). Then
it blew every couple of hours. The solution is Radius 2.0.1. I have had
it running for a couple of weeks and my monitoring script, the one that
restarts it upon failure, is also set to email me on each restart of
Radius: I have been running clean since the upgrade (no restarts). The
primary symptom for this (according to Livingston's voice Tech Support) is
an entry in the syslog file: died on signal 100. What it is trying to do
is kill off it's orphaned children. The parent mistakes itself for a
child and recklessly kills itself!
Radius 2.0.1 is the answer.
Tres
tres@chaffee.net
--Make it idiot proof and someone will make a better idiot.