Today I have 2 freeradius servers running... Both of them in a CentOS 5.6. 
The first is stable, without problems running freeradius 1.1.3.

The second is running freeradius 2.1.7 and in the last 3 months became very unstable.
After some time running the threads just shutdown... 
Reading the /var/log/radius/radius.log, the only message in the moment of the problem is:
Sun Jul  3 06:53:41 2011 : Info: Exiting normally.

When I check the running processes, the radiusd is running... the 'service radiusd status' command displays the pid of the runnind daemon... but radiusd is not listening the network ports anymore.

I tried to keep the radius in debug mode (radiusd -XXXX) for a week, but in this case the problem didn't happen.

Is there some way to force the radiusd print why it is exiting 'normally' ??

Thanks

--
Carlos Eduardo Tavares Terra
Red Hat Certified Engineer
Consultor em Infraestrutura de TI
GNU/Linux #413291 [http://counter.li.org]