view release on metacpan or  search on metacpan

lib/AC/MrGamoo/Job/Plan.pm  view on Meta::CPAN

# Function: 
#
# $Id: Plan.pm,v 1.1 2010/11/01 18:41:56 jaw Exp $

package AC::MrGamoo::Job::Plan;
use AC::MrGamoo::Debug 'plan';
use AC::Misc;

use strict;

my $REDUCEFACTOR = 1.9;		# QQQ - config?
my $MAPTARGETMIN = 8;		# try to have at least this many maps/server
my $MAPSIZELIMIT = 100_000_000;

sub new {
    my $class   = shift;
    my $job     = shift;
    my $servers = shift;
    my $files   = shift;

    return unless @$servers;

    # how many reduces?
    my $nr = _number_of_reduces( $job->{options}, scalar @$servers );

    # map servers to reduce bins
    my $redbin = _pick_reduce_bins( $nr, $servers );

    # plan out the map phase
    my @phase = 'map';
    my($planmap, $plancopy) = _plan_map( $job, $servers, $files, $nr, $redbin );
    my @task  = { phase => 'map', task => $planmap };

    # plan out the reduce phases
    my $nrp = @{$job->{mr}{content}{reduce}};
    for my $r (0 .. $nrp - 1){
        push @phase, "reduce/$r";
        # last reduce has 1 outfile, otherwise nr.
        my $nout = ($r == $nrp - 1) ? 1 : $nr;
        push @task,  { phase => "reduce/$r", task => _plan_reduce($job, $r, $nout, $redbin, $task[-1]{task}) };
    }

    # plan out a final phase
    if( $job->{mr}{content}{final} ){
        push @phase, 'final';
        push @task,  { phase => 'final', task => _plan_final($job, $redbin, $task[-1]{task}) };
    }

    # summary
    my %task;
    for my $ts (@task){
        for my $t ( @{$ts->{task}} ){
            $task{ $t->{id} } = $t;
        }
    }

    # debug("plan: " . dumper( \@task ));

    debug("infiles: " . @$files . ", precopy: " . @$plancopy . ", maps: " . @$planmap . ", reduces: $nr x $nrp");

    return bless {
        nserver		=> scalar(@$servers),
        nreduce		=> $nr,
        copying		=> $plancopy,
        phases		=> \@phase,
        taskplan	=> \@task,
        redbin		=> $redbin,
        taskidx		=> \%task,
    }, $class;
}

sub _number_of_reduces {
    my $config  = shift;
    my $nserver = shift;

    my $nr = $config->{reduces} + 0;
    $nr ||= int $nserver * $REDUCEFACTOR;
    $nr = 1 if $nr < 1;

    return $nr;
}

sub _pick_reduce_bins {
    my $nr      = shift;
    my $servers = shift;


    my @redbin;
    for my $bin (0 .. $nr-1){
        $redbin[$bin][0] = $servers->[ $bin % @$servers ]->{id};

        # pick alt location
        next unless @$servers > 1;
        $redbin[$bin][1] = $servers->[ ($bin + 1) % @$servers ]->{id};
    }
    shuffle(\@redbin);

    return \@redbin;
}

sub _plan_map {
    my $job     = shift;
    my $servers = shift;
    my $files   = shift;
    my $nr      = shift;
    my $redbin  = shift;

    # plan map
    #  divy files among servers
    #  split server + files into tasks

    my( $filemap, $copies ) = _plan_divy_files( $job, $files, $servers );

    my @maptask;
    for my $s (keys %$filemap){
        my $totalsize = 0;
        $totalsize += $_->{size} for @{$filemap->{$s}};;
        my $sizelimit = $totalsize / $MAPTARGETMIN;
        $sizelimit = $MAPSIZELIMIT if $sizelimit > $MAPSIZELIMIT;

        my @todo = sort { $b->{size} <=> $a->{size} } @{$filemap->{$s}};