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 <http://www-unix.mcs.anl.gov/mpi/mpich2/>. 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, "<alexandre.masselot at
   genebio.com>"

BUGS
   Please report any bugs or feature requests to "bug-parallel-mpich-mpd at
   rt.cpan.org", or through the web interface at
   <http://rt.cpan.org/NoAuth/ReportBug.html?Queue=Parallel-Mpich-MPD>. 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
       <http://annocpan.org/dist/Parallel-Mpich-MPD>

   * CPAN Ratings
       <http://cpanratings.perl.org/d/Parallel-Mpich-MPD>

   * RT: CPAN's request tracker
       <http://rt.cpan.org/NoAuth/Bugs.html?Dist=Parallel-Mpich-MPD>

   * Search CPAN
       <http://search.cpan.org/dist/Parallel-Mpich-MPD>

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.