• Proc::PID_File - takes care of PID file use for you.
• Proc::PID::File - OO way to do the above
• Linux::PID - for managing multi-threaded applications' PPID's on Linux (you still have to deal with files yourself).

#!/bin/bash

if (ps ax |grep -q "my_perl_script_name") ; then exit; fi
my_perl_script_name
[download]

I make my scripts write out their PID to a file when they start after they have checked they are unique. The check consists of seeing if the pid file exists, reading it and checking if the process on that PID looks like an instance of my script (just in case the OS re-used the pid for something else) clean exit will clear up the pid file. Here is the check...

# See if I am already running
# $my_name was derived earlier from $0
# $pidfile is set up in the scripts headers
if (-f $pidfile) {
    open PID, $pidfile;
    my $pid = <PID>;
    close PID;
    print "read pid $pid\n" if $debug;
    open PROCS, "ps -ef |";
    <PROCS>;    # loose the header
    while (<PROCS>) {
        (undef, my $this_pid) = split;
        if ($this_pid == $pid) {
            print "pid in use $_\n";
            # check this realy is us, not pid re-use by OS
            last unless /$my_name/;
            die "looks like an instance is already running, exiting\n"
+;
        }
    }
}
open PID, ">$pidfile";
print PID $$;
close PID;
[download]

That is the framework I use, it is a cut and paste from a larger script so will not run in isolation, sorry it is Friday and I want to go home so I can pack for my sailing holiday that starts tomorrow SEG

Cheers,
R.

That is very cool. Not to pester but would you give a snippet (or point to one) showing it in action?

I can't find my working example right now, but from memory, something like this:

#!/usr/bin/perl
use Fcntl qw(LOCK_EX LOCK_NB);
flock DATA, LOCK_EX | LOCK_NB or die "Already executing!";
... rest of code ...
# do not remove this __END__!
__END__
[download]

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.

Otherwise, what's wrong with using something like Cache::File? First thing you do is check the cache to see if the lock is set. If it is, leave. If it's not, set the lock. When you're done, release the lock.

$fh1 = &file_lock("$folder/index.html");
&file_unlock($fh1);

sub file_lock {
    my $fhandle;
    my $fpath = (shift) . '.lock';
    open($fhandle, ">$fpath");
    flock($fhandle, 2);
    return [$fhandle, $fpath];
}

sub file_unlock {
    my $inp = shift;
    my ($fhandle, $fpath) = @$inp;
    flock($fhandle, 8);
    close($fhandle);
    unlink($fpath);
}
[download]