Question

I’ve been working on a perl daemon for linux and here is its skeleton:

#!/usr/bin/env perl
use File::Copy;
use Socket;
use Sys::Hostname;
use POSIX qw(setsid);
use Env;

use Sys::Info::Constants qw( :device_cpu );

my $daemonName = 'proc';

my $proc;
my $error;
my $file = "Proc.pl";
my $pidfile = ">/var/run/proc.pid";
my $pid2check = "/var/run/proc.pid";
my $pid;


if (!$error) {
    LogMessage("$daemonName  : PID $proc : Begin");
}

if (!$error) {
    LogMessage("$daemonName  : PID $proc : Writing pid information to $pidfile");
    print FILE $proc . "\n";
    close (FILE);
}

$SLEEP_TIME = 5; # seconds

#Main loop of Daemon
while (!$error) {
    sleep($SLEEP_TIME);

}


if ($error) {
    LogMessage("$file : PID $proc : Error $error");
}

LogMessage("$file : PID $proc : END");

exit(0);


sub Daemonize {

if (!(chdir '/')) {
    $error .= "Can't chdir to /: $!";
}
if (!(umask 0)) {
    $error .= "Unable to umask 0";
}

unless (open STDIN, '/dev/null') {
    $error .= "Can't read /dev/null: $!";
}

open(OLD_OUT,">&STDOUT");  

#All print statments will now be sent to our log file
unless (open STDOUT, '>>/var/log/proc.log') {
    $error .= "Can't read /var/log/proc.log: $!";
}
#All error messages will now be sent to our log file
unless (open STDERR, '>>/var/log/proc.log') {
    $error .= "Can't write to /var/log/proc.log: $!";
}

defined($pid = fork);
#Exit if $pid exists (parent)

if ($pid)
{
    print OLD_OUT "Service successfully installed.\n";
    exit(0);
}

#As Child
setsid();
$proc = $$;
return ($proc);
 }


#Prints log messages
sub LogMessage {
my $message = $_[0];
print localtime() . " $message\n";
}

I’m relatively new to perl, and I was wondering what would be the easiest way to start and stop this daemon? Also how would I check to make sure there is not an instance of it already running?

Answer 1

Generally speaking, daemonisation is done with the double fork method. This is common idiom and there are libraries that do this for you which you should consider using. A pid or lockfile is often used to make sure that only one instance runs at a time.

If you’re planning to use this as a system process, it is also possible to use something like upstart or supervisord to do the process management and “only one at time” thing for you.