Hmm.. to throw some data into the mess:
I am running Livingston's pre-compiled radiusd v2.0.1 on BSDi's BSD/OS
v2.1 with all patches as my main radius server.
lithium@maslow:~>ps auxw | grep radiusd
root 223 0.0 0.5 348 148 co- I Mon07PM 0:09.81
/usr/libexec/radiusd -d /etc/raddb -a /var/log/radacct
root 227 0.0 0.6 320 188 co- I Mon07PM 0:11.90
/usr/libexec/radiusd -d /etc/raddb -a /var/log/radacct
So it has been running for about two hours shy of two days. I have 5
DEFAULT entries which combined are 33 lines worth. I am using the plant
text users file with one additional non-DEFAULT entry at the top.
According to logs the number of logins authenicated exceeds 1,000 in that
time period.
Killing and restarting the radius daemon yeilds these results:
root 5832 0.3 0.5 336 160 p0 S 6:11PM 0:00.02
/usr/libexec/radiusd -d /etc/raddb -a /var/log/radacct
root 5833 0.3 0.6 320 192 p0 S 6:11PM 0:00.01
/usr/libexec/radiusd -d /etc/raddb -a /var/log/radacct
The only difference besides less CPU time being used on the second one is
that the daemon is slighly larger the second time.
Am I missing something?
On Wed, 13 Aug 1997, Steven P. Crain wrote:
> > How large is it?
> > Can you tell me how many DEFAULT entries can me some memory problems?
>
> If you have more than one DEFAULT entry, the leak is significant. Take
> the size of all of your DEFAULT entries and divide by 2 for a very rough
> approximation of how much memory leaks per time a person attempts to log
> in.