Beefy Boxes and Bandwidth Generously Provided by pair Networks
There's more than one way to do things

Re: Ensuring only one copy of a perl script is running at a time

by benizi (Hermit)
on Dec 20, 2006 at 00:21 UTC ( #590826=note: print w/replies, xml ) Need Help??

in reply to Ensuring only one copy of a perl script is running at a time

Not sure if it's relevant to your situation, but this won't work over (some?) network filesystems. In my case, I have a home directory in AFS. Using either $0 or DATA flock'ing, it prevents two scripts on the same machine from running the same script. But, a script run from another machine proceeds as if nothing's flock'ed. I could see this being an issue if, say, a pool of webservers was serving your files.

A more robust solution might be to use a relational database system's locking mechanism. (This is often pretty convenient if you're already using a DB for other tasks.) My favorite method is to create a lock table as follows:

the MySQL:

create table foolock ( /* since we're exclusive per-program, make program primary */ program varchar(64) primary key, client varchar(255) not null, time timestamp );

and the Perl

#!/usr/bin/perl use strict; use warnings; use File::Basename; use Sys::Hostname; use DBI; my $dbh = DBI->connect('dbi:mysql:test:host','foolocker','foopass',{RaiseErr +or=>1}); my $lock_interval = 10; # minimum wait in milliseconds between attempt +s my $max_attempts = 5; # maximum number of attempts sub prog_client { ((basename $0), hostname.":$$"); } sub get_lock { sql_lock('insert into foolock (program, client) values (?,?)', pro +g_client); } sub finish_up { sql_lock('delete from foolock where program = ? and client = ?', p +rog_client); } sub sql_lock { my ($sql, @params) = @_; my $lock = $dbh->prepare_cached($sql); my $backoff = $lock_interval / 1000; for (1..$max_attempts) { my $got = eval { $dbh->do('lock tables foolock write'); $lock->execute(@params) >= 0; }; $dbh->do('unlock tables'); return 1 if $got; select undef, undef, undef, $backoff; $backoff *= 2; } return 0; } if (get_lock) { # do your thing finish_up; } else { # take appropriate action }

The get_lock and finish_up allow you to detect (via entries in the foolock table) when a script died without finishing. But, in the way it's laid out above, this prevents other processes from then acquiring locks. (Which for my task was desirable.)

Log In?

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://590826]
and all is quiet...

How do I use this? | Other CB clients
Other Users?
Others having an uproarious good time at the Monastery: (2)
As of 2017-04-25 01:32 GMT
Find Nodes?
    Voting Booth?
    I'm a fool:

    Results (447 votes). Check out past polls.