The init script of openvpn reads: stop() { ebegin "Stopping ${myservice}" start-stop-daemon --stop --exec /usr/sbin/openvpn --pidfile "${VPNPID}" eend $? } The start-stop-daemon command features either --exec or --pidfile. This is bad, the start-stop-daemon gives --exec more priority (http://www.koders.com/c/fid3ADAC80CB671F5EF100C7F7D63AFB1470A9C436B.aspx), so the script always tries to stop the OpenVPN using --exec method. Every stop command should have only one type of --exec or --pidfile with combination of other parameters. OpenVPN should be stopped via --pidfile method, it should be: start-stop-daemon --stop --pidfile "${VPNPID}" And here is the reason: init and start an eth up, start the openvpn, now down the eth, the openvpn process will go to stopped state. Now try to stop the openvpn with the init.d script. Youll get: No /usr/sbin/openvpn found running; none killed. With the pid method everything works fine. I have no clue why. The stopping via PID is the preffered method. The solution is to change this one line.
> Assignee: secure-tunneling@gentoo.org did not match anything Please create a proper alias.
Do you get the same error with baselayout-1.12.0_pre17-r2?
Its a bug in the ebuild.
(In reply to comment #3) > Its a bug in the ebuild. Really? You won't know until you try baselayout-1.12.0_pre17 (hint, it may just have changed start-stop-daemon)
I am sorry but this is not stable. I have 1.11.
We're in the process of making 1.12 stable.
