#!/usr/bin/perl -w
# vim: sts=4 sw=4 ts=8

# Munin markers:
#%# family=auto
#%# capabilities=autoconf suggest

# Author: Michael Renner <michael.renner@amd.co.at>

# Version: 0.0.5, 2009-05-22



=head1 NAME

linux_diskstat_ - Munin plugin to monitor various values provided
via C</proc/diskstats>

=head1 APPLICABLE SYSTEMS

Linux 2.6 systems with extended block device statistics enabled.


=head1 INTERPRETATION

Among the more self-describing or well-known values like C<throughput>
(Bytes per second) there are a few which might need further introduction.


=head2 Device Utilization

Linux provides a counter which increments in a millisecond-interval for as long
as there are outstanding I/O requests. If this counter is close to 1000msec
in a given 1 second timeframe the device is nearly 100% saturated. This plugin
provides values averaged over a 5 minute time frame per default, so it can't
catch short-lived saturations, but it'll give a nice trend for semi-uniform
load patterns as they're expected in most server or multi-user environments.


=head2 Device IO Time

The C<Device IO Time> takes the counter described under C<Device Utilization>
and divides it by the number of I/Os that happened in the given time frame,
resulting in an average time per I/O on the block-device level.

This value can give you a good comparison base amongst different controllers,
storage subsystems and disks for similar workloads.


=head2 Syscall Wait Time

These values describe the average time it takes between an application issuing
a syscall resulting in a hit to a blockdevice to the syscall returning to the
application.

The values are bound to be higher (at least for read requests) than the time
it takes the device itself to fulfill the requests, since calling overhead,
queuing times and probably a dozen other things are included in those times.

These are the values to watch out for when an user complains that C<the disks
are too slow!>.


=head3 What causes a block device hit?

A non-exhaustive list:

=over

=item * Reads from files when the given range is not in the page cache or the O_DIRECT
flag is set.

=item * Writes to files if O_DIRECT or O_SYNC is set or sys.vm.dirty_(background_)ratio
is exceeded.

=item * Filesystem metadata operations (stat(2), getdents(2), file creation,
modification of any of the values returned by stat(2), etc.)

=item * The pdflush daemon writing out dirtied pages

=item * (f)sync

=item * Swapping

=item * raw device I/O (mkfs, dd, etc.)

=back

=head1 ACKNOWLEDGEMENTS

The core logic of this script is based on the B<iostat> tool of the B<sysstat>
package written and maintained by Sebastien Godard.

=head1 SEE ALSO

See C<Documentation/iostats.txt> in your Linux source tree for further information
about the C<numbers> involved in this module.

L<http://www.westnet.com/~gsmith/content/linux-pdflush.htm> has a nice writeup
about the pdflush daemon.

=head1 AUTHOR

Michael Renner <michael.renner@amd.co.at>

=head1 LICENSE

GPLv2


=cut


use strict;


use File::Basename;
use Carp;
use POSIX;

# We load our own version of save/restore_state if Munin::Plugin is unavailable.
# Don't try this at home
eval { require Munin::Plugin; Munin::Plugin->import; };

if ($@) {
    fake_munin_plugin();
}


# Sanity check to ensure that the script is called the correct name.

if (basename($0) !~ /^linux_diskstat_/) {
    die qq(Please ensure that the name of the script and it's symlinks starts with "linux_diskstat_"\n);
}


############
# autoconf #
############

if ( defined $ARGV[0] && $ARGV[0] eq 'autoconf' ) {
    my %stats;

    # Capture any croaks on the way
    eval { %stats = parse_diskstats() };

    if ( !$@ && keys %stats ) {

        print "yes\n";
    }
    else {
        print "no\n";
    }
    exit 0;
}


###########
# suggest #
###########

if ( defined $ARGV[0] && $ARGV[0] eq 'suggest' ) {

    my %diskstats = parse_diskstats();

    my %suggested_devices;

  DEVICE:
    for my $devname ( sort keys %diskstats ) {

        # Skip devices without traffic
        next
          if ( $diskstats{$devname}->{'rd_ios'} == 0
            && $diskstats{$devname}->{'wr_ios'} == 0 );

        for my $existing_device ( @{ $suggested_devices{'iops'} } ) {

            # Filter out devices (partitions) which are matched by existing ones
            # e.g. sda1 -> sda, c0d0p1 -> c0d0
            next DEVICE if ( $devname =~ m/$existing_device/ );
        }

        push @{ $suggested_devices{'iops'} },       $devname;
        push @{ $suggested_devices{'throughput'} }, $devname;

        # Only suggest latency graphs if the device supports it
        if (   $diskstats{$devname}->{'rd_ticks'} > 0
            || $diskstats{$devname}->{'wr_ticks'} > 0 )
        {
            push @{ $suggested_devices{'latency'} }, $devname;
        }
    }

    for my $mode ( keys %suggested_devices ) {
        for my $device ( sort @{ $suggested_devices{$mode} } ) {

            my $printdev = translate_device_name($device, 'TO_FS');
            print "${mode}_$printdev\n";
        }
    }

    exit 0;
}


# Reading the scripts invocation name and setting some parameters,
# needed from here on

my $basename = basename($0);
my ( $mode, $device ) = $basename =~ m/linux_diskstat_(\w+)_([-+\w]+)$/;

if ( not defined $device ) {
    croak "Didn't get a device name. Aborting\n";
}

$device = translate_device_name($device, 'FROM_FS');

##########
# config #
##########

if ( defined $ARGV[0] && $ARGV[0] eq 'config' ) {

    my $pretty_device = $device;

    if ($device =~ /^dm-\d+$/) {
	$pretty_device = translate_devicemapper_name($device);
    }

    if ( $mode eq 'latency' ) {

        print <<EOF;
graph_title Disk latency for /dev/$pretty_device
graph_args --base 1000
graph_category disk

util.label Device utilization (percent)
util.type GAUGE
util.info Utilization of the device. If the time spent for I/O is close to 1000msec for a given second, the device is nearly 100% saturated.
util.min 0
svctm.label Average device IO time (ms)
svctm.type GAUGE
svctm.info Average time an I/O takes on the block device
svctm.min 0
avgwait.label Average IO Wait time (ms)
avgwait.type GAUGE
avgwait.info Average wait time for an I/O from request start to finish (includes queue times et al)
avgwait.min 0
avgrdwait.label Average Read IO Wait time (ms)
avgrdwait.type GAUGE
avgrdwait.info Average wait time for a read I/O from request start to finish (includes queue times et al)
avgrdwait.min 0
avgwrwait.label Average Write IO Wait time (ms)
avgwrwait.type GAUGE
avgwrwait.info Average wait time for a write I/O from request start to finish (includes queue times et al)
avgwrwait.min 0

EOF

    }
    elsif ( $mode eq 'throughput' ) {

        print <<EOF;
graph_title Disk throughput for /dev/$pretty_device
graph_args --base 1024
graph_vlabel Bytes/second
graph_category disk

rdbytes.label Read Bytes
rdbytes.type GAUGE
rdbytes.min 0
wrbytes.label Write Bytes
wrbytes.type GAUGE
wrbytes.min 0

EOF
    }
    elsif ( $mode eq 'iops' ) {

        print <<EOF;
graph_title Disk IOs for /dev/$pretty_device
graph_args --base 1000
graph_vlabel Units/second
graph_category disk

rdio.label Read IO/sec
rdio.type GAUGE
rdio.min 0
wrio.label Write IO/sec
wrio.type GAUGE
wrio.min 0
avgrqsz.label Average Request Size (KiB)
avgrqsz.type GAUGE
avgrqsz.min 0
avgrdrqsz.label Average Read Request Size (KiB)
avgrdrqsz.type GAUGE
avgrdrqsz.min 0
avgwrrqsz.label Average Write Request Size (KiB)
avgwrrqsz.type GAUGE
avgwrrqsz.min 0

EOF

    }
    else {
        croak "Unknown mode $mode\n";
    }
    exit 0;
}


########
# MAIN #
########


my %cur_diskstat = fetch_device_counters($device);


my ( $prev_time, %prev_diskstat ) = restore_state();

save_state( time(), %cur_diskstat );

# Probably the first run for the given device, we need state to do our job,
# so let's wait for the next run.
exit if ( not defined $prev_time or not %prev_diskstat );

calculate_and_print_values( $prev_time, \%prev_diskstat, \%cur_diskstat );



########
# SUBS #
########

sub calculate_and_print_values {
    my ( $prev_time, $prev_stats, $cur_stats ) = @_;

    my $bytes_per_sector = 512;

    my $interval = time() - $prev_time;

    my $read_ios  = $cur_stats->{'rd_ios'} - $prev_stats->{'rd_ios'};
    my $write_ios = $cur_stats->{'wr_ios'} - $prev_stats->{'wr_ios'};

    my $rd_ticks = $cur_stats->{'rd_ticks'} - $prev_stats->{'rd_ticks'};
    my $wr_ticks = $cur_stats->{'wr_ticks'} - $prev_stats->{'wr_ticks'};

    my $rd_sectors = $cur_stats->{'rd_sectors'} - $prev_stats->{'rd_sectors'};
    my $wr_sectors = $cur_stats->{'wr_sectors'} - $prev_stats->{'wr_sectors'};

    my $tot_ticks = $cur_stats->{'tot_ticks'} - $prev_stats->{'tot_ticks'};


    my $read_io_per_sec  = $read_ios / $interval;
    my $write_io_per_sec = $write_ios / $interval;

    my $read_bytes_per_sec  = $rd_sectors / $interval * $bytes_per_sector;
    my $write_bytes_per_sec = $wr_sectors / $interval * $bytes_per_sector;


    my $total_ios         = $read_ios + $write_ios;
    my $total_ios_per_sec = $total_ios / $interval;

    # Utilization - or "how busy is the device"?
    # If the time spent for I/O was close to 1000msec for
    # a given second, the device is nearly 100% saturated.
    my $utilization = $tot_ticks / $interval;

    # Average time an I/O takes on the block device
    my $servicetime =
      $total_ios_per_sec ? $utilization / $total_ios_per_sec : 0;

    # Average wait time for an I/O from start to finish
    # (includes queue times et al)
    my $average_wait = $total_ios ? ( $rd_ticks + $wr_ticks ) / $total_ios : 0;
    my $average_rd_wait = $read_ios  ? $rd_ticks / $read_ios  : 0;
    my $average_wr_wait = $write_ios ? $wr_ticks / $write_ios : 0;

    my $average_rq_size_in_kb =
      $total_ios
      ? ( $rd_sectors + $wr_sectors ) * $bytes_per_sector / 1024 / $total_ios
      : 0;
    my $average_rd_rq_size_in_kb =
      $read_ios ? $rd_sectors * $bytes_per_sector / 1024 / $read_ios : 0;
    my $average_wr_rq_size_in_kb =
      $write_ios ? $wr_sectors * $bytes_per_sector / 1024 / $write_ios : 0;

    my $util_print = $utilization / 10;


    if ( $mode eq 'latency' ) {
        print <<EOF;

util.value $util_print
svctm.value $servicetime
avgwait.value $average_wait
avgrdwait.value $average_rd_wait
avgwrwait.value $average_wr_wait

EOF
    }
    elsif ( $mode eq 'throughput' ) {

        print <<EOF;

rdbytes.value $read_bytes_per_sec
wrbytes.value $write_bytes_per_sec

EOF
    }
    elsif ( $mode eq 'iops' ) {

        print <<EOF;

rdio.value $read_io_per_sec
wrio.value $write_io_per_sec
avgrqsz.value $average_rq_size_in_kb
avgrdrqsz.value $average_rd_rq_size_in_kb
avgwrrqsz.value $average_wr_rq_size_in_kb

EOF

    }
    else {
        croak "Unknown mode $mode\n";
    }

}

sub read_diskstats {

    open STAT, '< /proc/diskstats'
      or croak "Failed to open '/proc/diskstats': $!\n";

    my @lines;

    for my $line (<STAT>) {

        # Strip trailing newline and leading whitespace
        chomp $line;
        $line =~ s/^\s+//;

        my @elems = split /\s+/, $line;

        # We explicitly don't support old-style diskstats
        # There are situations where only _some_ lines (e.g.
        # partitions on older 2.6 kernels) have fewer stats
        # numbers, therefore we'll skip them silently
        if ( @elems != 14 ) {
            next;
        }
        push @lines, \@elems;
    }

    close STAT or croak "Failed to close '/proc/diskstats': $!";
    return @lines;
}

sub read_sysfs {

    my ($want_device) = @_;

    my @devices;
    my @lines;

    if ( defined $want_device ) {

        # sysfs uses '!' as replacement for '/', e.g. cciss!c0d0
        $want_device =~ tr#/#!#;
        @devices = $want_device;
    }
    else {
        @devices = glob "/sys/block/*/stat";
        @devices = map { m!/sys/block/([^/]+)/stat! } @devices;
    }


    for my $cur_device (@devices) {
        my $stats_file = "/sys/block/$cur_device/stat";

        open STAT, "< $stats_file"
          or croak "Failed to open '$stats_file': $!\n";

        my $line = <STAT>;

        # Trimming whitespace
        $line =~ s/^\s+//;
        chomp $line;

        my @elems = split /\s+/, $line;

        croak "'$stats_file' doesn't contain exactly 11 or 15 values. Aborting"
          if ( @elems != 11 && @elems != 15 );

        # Translate the devicename back before storing the information
        $cur_device =~ tr#!#/#;

        # Faking missing diskstats values
        unshift @elems, ( -1, -1, $cur_device );

        push @lines, \@elems;

        close STAT or croak "Failed to close '$stats_file': $!\n";
    }

    return @lines;
}


sub parse_diskstats {

    my ($want_device) = @_;

    my @stats;

    if ( glob "/sys/block/*/stat" ) {

        @stats = read_sysfs($want_device);
    }
    else {
        @stats = read_diskstats();
    }

    my %diskstats;

    for my $entry (@stats) {

        my %devstat;

        # Hash-Slicing for fun and profit
        @devstat{
            qw(major minor devname
              rd_ios rd_merges rd_sectors rd_ticks
              wr_ios wr_merges wr_sectors wr_ticks
              ios_in_prog tot_ticks rq_ticks)
          }
          = @{$entry};

        $diskstats{ $devstat{'devname'} } = \%devstat;
    }

    return %diskstats;
}

sub fetch_device_counters {

    my ($want_device) = @_;

    my %diskstats = parse_diskstats($want_device);

    for my $devname ( keys %diskstats ) {

        if ( $want_device eq $devname ) {
            return %{ $diskstats{$devname} };
        }
    }
    return undef;
}


# We use '+' (and formerly '-') as placeholder for '/' in device-names
# used as calling name for the script.
sub translate_device_name {

    my ($device, $mode) = @_;

    if ($mode eq 'FROM_FS') {

	# Hackaround to mitigate issues with unwisely chosen former separator
	if ( not ($device =~ m/dm-\d+/)) {
	    $device =~ tr#-+#//#;
	}

    }
    elsif ($mode eq 'TO_FS') {

	$device =~ tr#/#+#;

    }
    else {
	croak "translate_device_name: Unknown mode\n";
    }

    return $device;
}


sub fake_munin_plugin {
    my $eval_code = <<'EOF';

use Storable;
my $storable_filename = basename($0);
$storable_filename = "/tmp/munin-state-$storable_filename";

sub save_state {
    my @state = @_;

    if ( not -e $storable_filename or -f $storable_filename ) {
	store \@state, $storable_filename or croak "Failed to persist state to '$storable_filename': $!\n";
    }
    else {
	croak "$storable_filename is probably not a regular file. Please delete it.\n";
    }
}

sub restore_state {

    if (-f $storable_filename) {
	my $state = retrieve($storable_filename);
	return @{$state};
    }
    else {
	return undef;
    }
}
EOF

    eval($eval_code);
}

sub translate_devicemapper_name {
        my ($device) = @_;

        my ($want_minor) = $device =~ m/^dm-(\d+)$/;

        croak "Failed to extract devicemapper id" unless defined ($want_minor);

	my $dm_major = find_devicemapper_major();
	croak "Failed to get device-mapper major number\n" unless defined $dm_major;

        for my $entry (glob "/dev/mapper/\*") {

                my $rdev = (stat($entry))[6];
                my $major = floor($rdev / 256);
                my $minor = $rdev % 256;

                if ($major == $dm_major && $minor == $want_minor) {

                        my $pretty_name = translate_lvm_name($entry);

                        return defined $pretty_name ? $pretty_name : $entry;

                }
        }
        # Return original string if the device can't be found.
        return $device;
}



sub translate_lvm_name {

        my ($entry) = @_;

        my $device_name = basename($entry);

        # Check for single-dash-occurrence to see if this could be a lvm devicemapper device.
        if ($device_name =~ m/(?<!-)-(?!-)/) {

                # split device name into vg and lv parts
                my ($vg, $lv) = split /(?<!-)-(?!-)/, $device_name, 2;
                return undef unless ( defined($vg) && defined($lv) );

                # remove extraneous dashes from vg and lv names
                $vg =~ s/--/-/g;
                $lv =~ s/--/-/g;

                $device_name = "$vg/$lv";

                # Sanity check - does the constructed device name exist?
                if (stat("/dev/$device_name")) {
                        return "$device_name";
                }

        }
        return undef;
}

sub find_devicemapper_major {

    open (FH, '< /proc/devices') or croak "Failed to open '/proc/devices': $!";

    my $dm_major;

    for my $line (<FH>) {
	chomp $line;

	my ($major, $name) = split /\s+/, $line, 2;

	next unless defined $name;

	if ($name eq 'device-mapper') {
	    $dm_major = $major;
	    last;
	}
    }
    close(FH);

    return $dm_major;
}