It starts OK but does not restart properly, because kill signal returns imediately, but shutdown of the deamon takes longer time and thus start attempt failed to bind on already opened socket

# service amavis-p0fanalyzer restart
Binding to socket [::1]:2345 failed (using IO::Socket::IP): Address already in use at /usr/local/sbin/p0f-analyzer.pl line 375.


p0fanalyzer_stop() should be changed to wait till daemon really die.

And should check if daemon is runnig and PID file exists, because second attempt to (re)start p0f emits this error:

# service amavis-p0fanalyzer restart
cat: /var/run/p0fanalyzer2.pid: No such file or directory
usage: kill [-s signal_name] pid ...
       kill -l [exit_status]
       kill -signal_name pid ...
       kill -signal_number pid ...
cat: /var/run/p0fanalyzer1.pid: No such file or directory
usage: kill [-s signal_name] pid ...
       kill -l [exit_status]
       kill -signal_name pid ...
       kill -signal_number pid ...


This error can be fixed by this simple change:

p0fanalyzer_stop()
{
    test -s ${amavis_p0fanalyzer_pidfile2} && /bin/kill `cat ${amavis_p0fanalyzer_pidfile2}` && rm ${amavis_p0fanalyzer_pidfile2}
    test -s ${amavis_p0fanalyzer_pidfile1} && /bin/kill `cat ${amavis_p0fanalyzer_pidfile1}` && rm ${amavis_p0fanalyzer_pidfile1}
}

I don't know how to handle the first problem with slowly dying daemon.

Also I am not sure if there must be the "rm" part in p0fanalyzer_stop() function, because man daemon says that the pid file will be removed.

cite: The program is executed in a
   spawned child process while the daemon waits until it terminates
   to keep the child_pidfile locked and removes it after the process
   exits.