package IPC::ConcurrencyLimit::Lock::Flock;
use 5.008001;
use strict;
use warnings;
use Carp qw(croak);
use File::Path qw();
use File::Spec;
use Fcntl qw(:DEFAULT :flock);
use IO::File ();
use IPC::ConcurrencyLimit::Lock;
our @ISA = qw(IPC::ConcurrencyLimit::Lock);
sub new {
my $class = shift;
my $opt = shift;
my $max_procs = $opt->{max_procs}
or croak("Need a 'max_procs' parameter");
my $path = $opt->{path}
or croak("Need a 'path' parameter");
my $lock_mode = lc($opt->{lock_mode} || 'exclusive');
if ($lock_mode !~ /^(?:exclusive|shared)$/) {
croak("Invalid lock mode '$lock_mode'");
}
my $self = bless {
max_procs => $max_procs,
path => $path,
lock_fh => undef,
lock_file => undef,
id => undef,
lock_mode => $lock_mode,
} => $class;
$self->_get_lock() or return undef;
return $self;
}
sub _get_lock {
my $self = shift;
File::Path::mkpath($self->{path});
my $lock_mode_flag = $self->{lock_mode} eq 'shared' ? LOCK_SH : LOCK_EX;
for my $worker (1 .. $self->{max_procs}) {
my $lock_file = File::Spec->catfile($self->{path}, "$worker.lock");
sysopen(my $fh, $lock_file, O_RDWR|O_CREAT)
or die "can't open '$lock_file': $!";
if (flock($fh, $lock_mode_flag|LOCK_NB)) {
$self->{lock_fh} = $fh;
seek($fh, 0, 0);
truncate($fh, 0);
print $fh $$;
$fh->flush;
$self->{id} = $worker;
$self->{lock_file} = $lock_file;
last;
}
close $fh;
}
return undef if not $self->{id};
return 1;
}
sub lock_file { $_[0]->{lock_file} }
sub path { $_[0]->{path} }
sub DESTROY {
my $self = shift;
# should be superfluous
close($self->{lock_fh}) if $self->{lock_fh};
}
1;
__END__
=head1 NAME
IPC::ConcurrencyLimit::Lock::Flock - flock() based locking
=head1 SYNOPSIS
use IPC::ConcurrencyLimit;
=head1 DESCRIPTION
This locking strategy implements C<flock()> based concurrency control.
Requires that your system has a sane C<flock()> implementation as well
as a non-blocking C<flock()> mode.
Inherits from L<IPC::LimitConcurrency::Lock>.
Take care not to attempt to use this on an NFS share or any other file
system that does not implement atomic C<flock()>!
=head1 METHODS
=head2 new
Given a hash ref with options, attempts to obtain a lock in
the pool. On success, returns the lock object, otherwise undef.
Required options:
=over 2
=item C<path>
The directory that will hold the lock files.
Created if it does not exist.
It is suggested not to use a directory that may hold other data.
=item C<max_procs>
The maximum no. of locks (and thus usually processes)
to allow at one time.
=back
Other options:
=over 2
=item C<lock_mode>
Defaults to C<exclusive> locks.
In particular circumstance, you might want to set this to C<shared>.
This subverts the way the normal concurrency limit works, but allows
entirely different use cases.
=back
=head2 lock_file
Returns the full path and name of the lock file.
=head2 path
Returns the directory in which the lock files resides.
=head1 AUTHOR
Steffen Mueller, C<smueller@cpan.org>
Yves Orton
=head1 ACKNOWLEDGMENT
This module was originally developed for booking.com.
With approval from booking.com, this module was generalized
and put on CPAN, for which the authors would like to express
their gratitude.
=head1 COPYRIGHT AND LICENSE
(C) 2011, 2012 Steffen Mueller. All rights reserved.
This code is available under the same license as Perl version
5.8.1 or higher.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
=cut