When I run /etc/init.d/radiusd start it won't work until I change the start function in the ebuild to: start() { # Comment out the following line to get faster startups #checkconfig || return 1 ebegin "Starting radiusd" echo `start-stop-daemon --start --quiet --exec /usr/sbin/radiusd -- ${RADIUSD_OPTS}` #>/dev/null eend $? } Two things to point out. First the checkconfig function is killed and the start-stop-daemon line has to be surrounded in backticks and prepended with the echo command. I have no idea why this works. I also thave >/dev/null commented out, but that's just a debug thing. I also have "-p1812" in /etc/conf.d/radius. This doesn't seem to be the case on everyones machine, but it has been a confirmed issue on some other machines. See the forum URL for details on that. The stop script does not work. I have not tried messing with it and have opted for using killall radiusd in the mean time. Reproducible: Sometimes Steps to Reproduce:
Oh, sorry. Forgot to mention the error that comes out when it isn't working. It complains: "Config not ok! (try /usr/sbin/check-radiusd-config )" in the logs and "auth bind: Address already in use There appears to be another RADIUS server already running on the authentication port UDP 32768." at the prompt The config script does not specify port 32768 and there is no running instance of radius at all, let alone on the port. When I remove the checkconfig from the startup script it just dies without notice.
did you run check-radiusd-config? freeradius isn't a program that you just run out of the box. you should first configure it.
I don't recall and unfortunately, the computer I was testing this out on had the processor fry from something (ps maybe). I am working my way out of a whole and can test it again on the replacement machine soon. Thank you for the tip on that radius config check command.
it is a configuration problem, which is already indicated in log file. the ebuild process cannot do more than it already does since freeradius is a highly customizable service.