NAME Parallel::Mpich::MPD - Mpich MPD wrapper *$VERSION* SYNOPSIS use Parallel::Mpich::MPD; # VERBOSE LEVEL #$Parallel::Mpich::MPD::Common::WARN=1; #$Parallel::Mpich::MPD::Common::DEBUG=1; #CHECK ENV Parallel::Mpich::MPD::Common::env_Hostsfile(/path/to/machinesfile); Parallel::Mpich::MPD::Common::env_MpichHome(/path/to/mpdhome); Parallel::Mpich::MPD::Common::env_Check(); #CHECK MPD AND NETWORK my %hostsup; my %hostsdown; my %info=Parallel::Mpich::MPD::info(); #check mpd master print Dumper(\%info) %hostsup= Parallel::Mpich::MPD::Common::checkHosts(hostsdown => \%hostsdown ); #check ping and ssh on machines %hostsup= Parallel::Mpich::MPD::check( reboot =>1:0, hostsdown=>\%hostsdown); #check mpds instances and try to repair ... # USE MPD Parallel::Mpich::MPD::boot(); #start mpd instances defined by default machinesfile my $alias1=Parallel::Mpich::MPD::makealias(); if Parallel::Mpich::MPD::createJob(cmd => $cmd, params => $parms, $machinesfile => $hostsfile, alias => $alias1)){ my $job=Parallel::Mpich::MPD::findJob(jobalias => $alias, getone => 1); $job->sig_kill() if defined $job; } DESCRIPTION This *Parallel::Mpich::MPD*, a wrapper module for MPICH2 Process Management toolkit from . The wrapper include the following tools: basic configuration, mpdcheck, mpdboot, mpdcleanup, mpdtrace, mpdringtest, mpdallexit, mpiexec, mpdsigjob and mpdlistjobs. EXPORT boot(hosts => @hosts, machinesfile => $machines, checkOnly => 1|0, output => \$output) starts a set of mpd's on a list of machines. boot try to verify that the hosts in the host file are up before attempting start mpds on any of them. rebootHost(host => $hostname) restart mpd on the specified host. rebootHost will kill old mpds before restarting a new one. The killed MPDS are filtered by specific port and host. check(machinesfile => $file, hostsup => \%hosts, hostsdown => \%hostsdown , reboot => 1) Check if MPD master and nodes are well up. If MPD master is down it try to ping and ssh machines. If you use the option reboot, check will try to restart mpd on specified machines or to reboot the master. info( ) return an %info of the master with the following keys (master, hostname, port) validateMachinesfile(machinefiles => $filename) check with mpdtrace if all machines specified by filename are up. If not, a temporary file is created with the resized machinesfile shutdown( ) causes all mpds in the ring to exit createJob({cmd => $cmd , machinesfile=> $filename, [params => $params], [ncpu => $ncpu], [alias => $alias]) start a new job with the command line and his params. It return true if ok. WARNING ncpu could be redefined if mpdtrace return à small hosts list Example: Parallel::Mpich::MPD::createJob(cmd => $cmd, params => $parms, ncpu => '3', alias => 'job1'); listJobs([mpdlistjobs_contents=>$str]) Return an Parallel::Mpich::MPD::Job array for all available jobs If mpdlistjobs_contents argument is present, the code will not call mpdlistjobs but take the parameter as a fake results of this command findJob([%criteria][, return=>(getone|host2pidlist)) find a job from crtiteria. It return a Job instance or undef for no match params Criteria can be of username=>'somename' or username=>\@arrayOfNames jobid=>'somename' or jobid=>\@arrayOfJobid jobalias=>'somename' or jobalias=>\@arrayOfJobalias To set an array of names; $criteria{psid} [&& $criteria{rank}] You can select psid from the specified rank. $criteria{reloadlist} force the call of listjobs return value By default (no return=>... argument), returned value will be a hash (or a hash ref, depending on context), {jobid1=>$job1, jobid2=>job2,...} return=>'getone' Will force to return the one job matching, or undef if none or error if many. return=>'host2pidlist' return a hash (or a ref to this hash, depending on context), host=>\@pidlist Examples trace([hosts => %hosts], long => 1) Lists the hostname of each of the mpds in the ring return true if ok [long=1] shows full hostnames and listening ports and ifhn makealias( ) "handle-" + PID + RAND(100) + Instance COUNTER++ return a uniq string alias clean([hosts => %hosts] [killcmd=>"cmd"]) Removes the Unix socket on local (the default) and remote machines This is useful in case the mpd crashed badly and did not remove it, which it normally does [hosts => %hosts] use specified hosts , [$cleancmd="cmd"] user defined kill command AUTHOR Olivier Evalet, Alexandre Masselot, "" BUGS Please report any bugs or feature requests to "bug-parallel-mpich-mpd at rt.cpan.org", or through the web interface at . I will be notified, and then you'll automatically be notified of progress on your bug as I make changes. SUPPORT You can find documentation for this module with the perldoc command. perldoc Parallel::Mpich::MPD You can also look for information at: * AnnoCPAN: Annotated CPAN documentation * CPAN Ratings * RT: CPAN's request tracker * Search CPAN ACKNOWLEDGEMENTS COPYRIGHT & LICENSE Copyright 2006 Olivier Evalet, Alexandre Masselot, all rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.