package TAP::Harness::Remote; our $VERSION = '1.10'; use warnings; use strict; use Carp; use base 'TAP::Harness'; use constant config_path => "$ENV{HOME}/.remote_test"; use File::Spec; use Cwd; use YAML; =head1 NAME TAP::Harness::Remote - Run tests on a remote server farm =head1 SYNOPSIS prove -l --state=save,slow --harness TAP::Harness::Remote t/*.t =head1 DESCRIPTION Sometimes you want to run tests on a remote testing machine, rather than your local development box. C allows you so reproduce entire directory trees on a remote server via C, and spawn the tests remotely. It also supports round-robin distribution of tests across multiple remote testing machines. =head1 USAGE C synchronizes local directories to the remote testing server. All tests that you wish to run remotely must be somewhere within these "local testing directories." You should configure this set by creating or editing your F<~/.remote_test> file: --- ssh: /usr/bin/ssh local: - /path/to/local/testing/root/ - /path/to/another/testing/root/ user: username host: remote.testing.host.example.com root: /where/to/place/local/root/on/remote/ perl: /usr/bin/perl master: 1 ssh_args: - -x - -S - '~/.ssh/master-%r@%h:%p' rsync_args: - -C - --exclude - blib env: FOO: bar See L for more details on the individual configuration options. Once your F<~/.remote_test> is configured, you can run your tests remotely, using: prove -l --harness TAP::Harness::Remote t/*.t Any paths in C<@INC> which point inside your local testing roots are rewritten to point to the equivilent path on the remote host. This is especially useful if you are testing a number of inter-related modules; by placing all of them all as local testing roots, and adding all of their C paths to your C, you can ensure that the remote machine always tests your combination of the modules, not whichever versions are installed on the remote host. If you have a farm of remote hosts, you may change the C configuration variable to be an array reference of hostnames. Tests will be distributed in a round-robin manner across the hosts. Each host will run as many tests in parallel as you specified with C<-j>. Especially when running tests in parallel, it is highly suggested that you use the standard L C<--state=save,slow> option, as this ensures that the slowest tests will run first, reducing your overall test run time. =head1 METHODS =head2 new Overrides L to load the local configuration, and add the necessary hooks for when tests are actually run. =cut sub new { my $class = shift; my $self = $class->SUPER::new(@_); $self->load_remote_config; for ( @{$self->remote_config("local")} ) { die "Local testing root ($_) doesn't exist\n" unless -d $_; } # Find which testing root we're under die "Current path isn't inside of local testing roots (@{$self->remote_config('local')})\n" unless defined $self->rewrite_path( Cwd::cwd ); die "Testing host not defined\n" unless grep { defined and not /\.example\.com$/ } @{ $self->remote_config("host") }; die "Can't find or execute ssh command: @{[$self->remote_config('ssh')]}\n" unless -e $self->remote_config("ssh") and -x $self->remote_config("ssh"); $ENV{HARNESS_PERL} = $self->remote_config("ssh"); $self->jobs( $self->jobs * @{ $self->remote_config("host") } ); $self->callback( before_runtests => sub { $self->setup(@_) } ); $self->callback( parser_args => sub { $self->change_switches(@_) } ); return $self; } =head2 config_path Returns the path to the configuration file; this is usually C<$ENV{HOME}/.remote_test>. =head2 default_config Returns, as a hashref, the default configuration. See L. =cut sub default_config { return { user => "smoker", host => "smoke-server.example.com", root => "/home/smoker/remote-test/$ENV{USER}/", perl => "/home/smoker/bin/perl", local => [ "$ENV{HOME}/remote-test/" ], ssh => "/usr/bin/ssh", ssh_args => [ "-x", "-S", "~/.ssh/master-%r@%h:%p" ], rsync_args => [ "-C" ], master => 1, env => {}, }; } =head2 load_remote_config Loads and canonicalizes the configuration. Writes and uses the default configuration (L) if the file does not exist. =cut sub load_remote_config { my $self = shift; unless ( -e $self->config_path and -r $self->config_path ) { YAML::DumpFile( $self->config_path, $self->default_config ); } $self->{remote_config} = YAML::LoadFile( $self->config_path ); # Make local path into an arrayref $self->{remote_config}{local} = [ $self->{remote_config}{local} ] unless ref $self->{remote_config}{local}; # Strip trailing slashes in local dirs, for rsync $self->{remote_config}{local} = [map {s|/$||; $_} @{$self->{remote_config}{local}}]; # Host should be an arrayref $self->{remote_config}{host} = [ $self->{remote_config}{host} ] unless ref $self->{remote_config}{host}; # Ditto ssh_args $self->{remote_config}{ssh_args} = [ split ' ', ( $self->{remote_config}{ssh_args} || "") ] unless ref $self->{remote_config}{ssh_args}; # Also, rsync_args $self->{remote_config}{rsync_args} = [ split ' ', ($self->{remote_config}{rsync_args} || "") ] unless ref $self->{remote_config}{rsync_args}; # Defaults for env $self->{env} ||= {}; } =head2 remote_config KEY Returns the configuration value set fo the given C. =cut sub remote_config { my $self = shift; $self->load_remote_config unless $self->{remote_config}; return $self->{remote_config}->{ shift @_ }; } =head2 userhost [HOST] Returns a valid C string; host is taken to be the first known host, unless provided. =cut sub userhost { my $self = shift; my $userhost = @_ ? shift : $self->remote_config("host")->[0]; $userhost = $self->remote_config("user") . "\@" . $userhost if $self->remote_config("user"); return $userhost; } =head2 start_masters Starts the ssh master connections, if support for them is enabled. Otherwise, does nothing. See the man page for C for more information about master connections. =cut sub start_masters { my $self = shift; return unless $self->remote_config("master"); local $SIG{USR1} = sub { die "Failed to set up SSH master connections\n"; }; my $parent = $$; for my $host ( @{ $self->remote_config("host") } ) { my $userhost = $self->userhost($host); my $pid = fork; die "Fork failed: $!" unless $pid >= 0; if ( not $pid ) { # Make sure we clean out this list, so we don't run # anything on _our_ DESTROY $self->{ssh_master} = {}; # Start the master system($self->remote_config("ssh"), @{ $self->remote_config("ssh_args") }, "-M", "-N", $userhost); # Signal the parent when we're done; we're still within 2 # seconds of starting, we'll catch this and abort. kill 'USR1', $parent; exit; } $self->{ssh_master}{$userhost} = $pid; } # During this sleep, we're waiting for our kids to tell us that # they died. sleep 5; } =head2 setup Starts the openssh master connections if need be (see L), then L's over the local roots. Additionally, stores a rewritten PERL5LIB path such that any directories which point into the local root are included in the remote PERL5LIB as well. =cut sub setup { my $self = shift; $SIG{USR1} = sub {}; $self->start_masters; $self->rsync; # Set up our perl5lib $self->{perl5lib} = join( ":", grep {defined} map {$self->rewrite_path($_)} split( /:/, $ENV{PERL5LIB} || "" ) ); $self->{perl5lib} =~ s/^(lib:){1,}/lib:/; # Also, any other env vars $self->{env} = []; for my $k (keys %{$self->remote_config("env")}) { my $val = $self->remote_config("env")->{$k}; $val =~ s/'/'"'"'/g; push @{$self->{env}}, "$k='$val'"; } } =head2 rsync Sends all local roots to the remote hosts, one at a time, using C. =cut sub rsync { my $self = shift; for my $host ( @{ $self->remote_config("host") } ) { my $userhost = $self->userhost($host); my $return = system( qw!rsync -avz --delete!, @{$self->remote_config('rsync_args')}, qq!--rsh!, $self->remote_config("ssh") . " @{$self->remote_config('ssh_args')}", @{$self->remote_config("local")}, "$userhost:" . $self->remote_config("root") ); die "rsync to $userhost failed" if $return; } } =head2 rewrite_path PATH Rewrites the given local C into the remote path on the testing server. Returns undef if the C isn't inside any of the configured local paths. =cut sub rewrite_path { my $self = shift; my $path = shift; my $remote = $self->remote_config("root"); for my $local ( @{$self->remote_config("local")} ) { if ($path =~ /^$local/) { $path =~ s{^$local}{$remote . "/" . (File::Spec->splitpath($local))[-1]}e; return $path; } } return undef; } =head2 DESTROY Tears down the ssh master connections, if they were started. =cut sub DESTROY { my $self = shift; return unless $self->remote_config("master"); for my $userhost ( keys %{ $self->{ssh_master} || {} } ) { next unless kill 0, $self->{ssh_master}{$userhost}; system $self->remote_config("ssh"), @{ $self->remote_config("ssh_args") }, "-O", "exit", $userhost; } } =head2 change_switches Changes the switches around, such that the remote perl is called, via ssh. This code is called once per test file. =cut sub change_switches { my ( $self, $args, $test ) = @_; my $remote = $self->remote_config("root"); my @other = grep { not /^-I/ } @{ $args->{switches} }; my @inc = map {"-I$_"} grep {defined $_} map { s/^-I//; $self->rewrite_path($_) } grep {/^-I/} @{ $args->{switches} }; my $host = $self->remote_config("host") ->[ $self->{hostno}++ % @{ $self->remote_config("host") } ]; my $userhost = $self->userhost($host); $args->{switches} = [ @{ $self->remote_config("ssh_args") }, $userhost, "cd", $self->rewrite_path( Cwd::cwd ), "&&", "PERL5LIB='@{[$self->{perl5lib}]}'", @{$self->{env}}, $self->remote_config("perl"), @other, @inc ]; } =head1 CONFIGURATION AND ENVIRONMENT Configuration is done via the file C<~/.remote_test>, which is a YAML file. Valid keys are: =over =item user The username to use on the remote connection. =item host The host to connect to. If this is an array reference, tests will be distributed, round-robin fashion, across all of the hosts. This does also incur the overhead of rsync'ing to each host. =item root The remote testing root. This is the place where the local roots will be C'd to. =item local The local testing roots. This can be either an array reference of multiple roots, or a single string. Files under each of these directories will be C'd to the remote server. All tests to be run remotely must be within these roots. =item perl The path to the C binary on the remote host. =item ssh The path to the local C binary. =item ssh_args Either a string or an array reference of arguments to pass to ssh. Suggested defaults include C<-x> and C<-S ~/.ssh/master-%r@%h:%p> =item master If a true value is given for this, will attempt to use OpenSSH master connections to reduce the overhead of making repeated connections to the remote host. =item rsync_args Either a string or an array reference of arguments to pass to rsync. You can use this, for say C<--exclude blib>. The arguments C<-avz --delete> are fixed, and then any C are appended. C<-C> is generally a useful and correct option, and is the default when creating new F<.remote_test> files. See L for more details. =item env A hash reference of environment variable names and values, to be used on the remote host. =back =head1 DEPENDENCIES A recent enough TAP::Harness build; 3.03 or later should suffice. Working copies of OpenSSH and rsync. =head1 BUGS AND LIMITATIONS Aborting tests using C<^C> may leave dangling processes on the remote host. Please report any bugs or feature requests to C, or through the web interface at L. =head1 AUTHOR Alex Vandiver C<< >> =head1 LICENCE AND COPYRIGHT Copyright (c) 2007-2008, Best Practical Solutions, LLC. All rights reserved. This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See L. =head1 DISCLAIMER OF WARRANTY BECAUSE THIS SOFTWARE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE SOFTWARE, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE SOFTWARE "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS WITH YOU. SHOULD THE SOFTWARE PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR, OR CORRECTION. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE SOFTWARE AS PERMITTED BY THE ABOVE LICENCE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE SOFTWARE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE SOFTWARE TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. =cut 1;