hadoop-dfs-plugin

#!/usr/bin/perl

#
# Description:
# This plugin shows basic filesystem information and statistics about the HDFS
# filesystem.
#
# Installation notes:
# Symlink this file to hadoop_hdfs_block or hadoop_hdfs_capacity to get block
# or capacity information about the DFS.
#
#
# Needs following minimal configuration in plugin-conf.d/munin-node:
# [hadoop_hdfs_block]
# user hdfs
#
# [hadoop_hdfs_capacity]
# user hdfs
#
# Author: KARASZI István <muninexchange@spam.raszi.hu>
#

use strict;
use File::Basename qw(basename);
use Munin::Plugin;


#
# main
#
my $type = &getType($0);
if ($ARGV[0]) {
	if ($ARGV[0] eq "autoconf") {
		print "yes\n";
	} elsif ($ARGV[0] eq "config") {
		&printConfig();
	} else {
		exit(1);
	}
} else {
	&getStatistics();
}

#
# methods
#
sub getType {
	my($command) = @_;

	my $scriptname = &File::Basename::basename($command);
	if ($scriptname =~ /_([^_]+)$/) {
		if (grep($1, ('block', 'capacity'))) {
			return $1;
		}
	}

	die("Invalid command type '$scriptname'");
}

sub printConfig {
	if ($type eq "block") {
		print <<EOT;
graph_title DFS Statistics
graph_category fs
graph_vlabel count
block.label Under replicated blocks
corrupt.label Blocks with corrupt replicas
missing.label Missing blocks
EOT
	} elsif ($type eq "capacity") {
		print <<EOT;
graph_title DFS Capacity
graph_category fs
graph_vlabel capacity
configured.label Configured
present.label Present
remaining.label DFS Remaining
EOT
	}
}

sub getBlock {
	my($line) = @_;

	if ($line =~ /^Under replicated blocks: (\d+)/) {
		printf("block.value %d\n", $1);
	} elsif ($line =~ /^Blocks with corrupt replicas: (\d+)/) {
		printf("corrupt.value %d\n", $1);
	} elsif ($line =~ /^Missing blocks: (\d+)/) {
		printf("missing.value %d\n", $1);
	}
}

sub getCapacity {
	my($line) = @_;

	if ($line =~ /^Configured Capacity: (\d+)/) {
		printf("configured.value %d\n", $1);
	} elsif ($line =~ /^Present Capacity: (\d+)/) {
		printf("present.value %d\n", $1);
	} elsif ($line =~ /^DFS Remaining: (\d+)/) {
		printf("remaining.value %d\n", $1);
	}
}

sub getStatistics {
	open(DFSADMIN, "hdfs dfsadmin -report|") || die("Cannot open dfsadmin: $!");
	while(defined(my $line = <DFSADMIN>)) {
		chomp($line);
                if ($line =~ /-------------------------------------------------/) {
                   last
                }

		if ($type eq "block") {
			&getBlock($line);
		} elsif ($type eq "capacity") {
			&getCapacity($line);
		}
	}
	close(DFSADMIN)
}