#!/usr/bin/env perl

use warnings;
use strict;
use Cwd;

# PERL script to do round-robin uploads to MSS at NCSA
# Usage is
# NCSAstore.pl <destDir> <nodeList> <fileList>
# where
# destDir  is the destination directory on MSS
# nodeList is the PBS_NODEFILE
# fileList is an "ls -1 --hide-control-chars" list of the
#          files to be stored

my( $destDir, $nodeList, $fileList );
my( $rcs, $rev, $rcsdate );
my( $currDir );
my( @files, @machines );
my( $numFiles, $numMachines );
my( $currFile, $currMachine );
my( $msscommand, $sshcommand );
my( $i, @processList, $childPID );



# Write a welcome

$rcs = '$RCSfile: NCSAstore.pl,v $';
$rev = '$Revision: 1.5 $';
$rcsdate = '$Date: 2006/02/17 20:55:56 $';

print "NCSAstore\n";
print "=========\n\n";
print "$rcs\n";
print "$rev\n";
print "$rcsdate\n\n";



# Fetch some arguments

$destDir = shift @ARGV || die "No destination directory!\n";
$nodeList = shift @ARGV || die "No node list!\n";
$fileList = shift @ARGV || die "No file lsit!\n";

$currDir = Cwd::getcwd;

# Slurp in the machines
open( MACHLIST, "<".$nodeList ) || die "Can't open list of machines!\n";
@machines= <MACHLIST>;
close( MACHLIST );
$numMachines = @machines;

# Slurp in the files
open( FILELIST, "<".$fileList ) || die "Can't open list of files!\n";
@files = <FILELIST>;
close( FILELIST );
$numFiles = @files;

print "We are going to attempt to store $numFiles files ";
print "using $numMachines machines\n\n";

# Remove CR/LF etc.
chomp @files;
chomp @machines;

# Loop

for( $i=0; $i<=$numFiles-1; $i++ )
{
    # Find the file and machine
    $currFile = $files[$i];
    $currMachine = $machines[$i % $numMachines];
    print "Storing $currFile using $currMachine\n";

    # Make up the MSS command
    $msscommand = '"'."cd $destDir , lcd $currDir, put $currFile".'"';
    
    # Make the SSH command
    $sshcommand = "ssh $currMachine msscmd $msscommand";

    # Fork off
    $childPID = fork();
    if( $childPID )
    {
	# This is the parent
	# Stash the PID away
	$processList[$i % $numMachines] = $childPID;
    }
    else
    {
	# This is the child
	# Run the ssh command after a pause to help
	# keep the output in order
	sleep 1;
	exec( "$sshcommand" );
    }


    # At the end of each pass through the machine list
    # wait for completion
    if(  $i % $numMachines == $numMachines-1 )
    {
	print "Waiting for completion of current batch: ";
	print "@processList\n\n";
	foreach $childPID ( @processList )
	{
	    waitpid( $childPID, 0 );
	    print "Child $childPID completed\n";
	}
	# Remember to nullify the process list
	@processList = ();
	print "Completed this batch\n\n";
    }
}

print "\n\nFinal wait for completion\n";
print "Remaining processes: @processList\n\n";
foreach $childPID ( @processList )
{
    waitpid( $childPID, 0 );
    print "Child $childPID completed\n";
}

print "\n\nNCSAstore complete\n\n";