BackupPC-Check/check_size_consistency.pl

#!/usr/bin/perl

use lib "/usr/share/BackupPC/lib";
use BackupPC::Lib;
use BackupPC::CGI::Lib;
use POSIX;
use JSON;
use Getopt::Long;
use Statistics::Descriptive;
use Data::Dumper;

my $host   = undef;
my $entity = undef;
my $pretty = 0;

GetOptions(
  "host=s"   => \$host,
  "entity=s" => \$entity,
  "pretty"   => \$pretty
);

# We need to switch to backuppc UID/GID
my $uid = getuid();
my $gid = getgid();
my (undef,undef,$bkpuid,$bkpgid) = getpwnam('backuppc');
setuid($bkpuid) if ($uid ne $bkpuid);
setgid($bkpgid) if ($gid ne $bkpgid);

my $bpc = BackupPC::Lib->new();
my $mainConf = $bpc->ConfigDataRead();
my $json = {};

if ( $host ) {
  my $hostConf = $bpc->ConfigDataRead($host);
  my $conf = { %$mainConf, %$hostConf };
  my $age = -1;
  $json = {
    bkp             => 0,
    full_size       => 0,
    errors          => 0,
    new_size        => 0,
    new_size_avg    => 0,
    new_size_median => 0,
    new_size_q1     => 0,
    new_size_q3     => 0,
    duration        => 0,
    comp_ratio      => 0
  };

  my $lastXferErrors        = 0;
  my $maxErrors             = 0;
  my $new_size_of_last_full = 0;
  my @bpc_info              = $bpc->BackupInfoRead($host);
  my $sizes                 = new Statistics::Descriptive::Full;

  if ( scalar( @bpc_info ) ){
    foreach my $backup ( @bpc_info ) {
      # Skip partial or active backups
      next if ( $backup->{type} !~ m/^full|incr$/ );
      if ( $backup->{type} eq "full" ) {
        $json->{full_size} = $backup->{size};
        $new_size_of_last_full = $backup->{sizeNew};
      }
      # Push all the sizes in our data set to compute avg sizes
      # Exclude backup N°0 as it'll always have much more new data than normal backups
      $sizes->add_data($backup->{sizeNew}) unless ( $backup->{num} == 0 );
      $json->{bkp}++;
    }

    # Ignore the last backup if it's not full or incr (which means it's either partial or active)
    my $i = ( $bpc_info[-1]->{type} =~ m/^full|incr$/ ) ? -1 : -2;

    $json->{errors}      = $bpc_info[$i]->{xferErrs};
    $json->{new_size}    = $bpc_info[$i]->{sizeNew};
    $json->{duration}    = $bpc_info[$i]->{endTime} - $bpc_info[$i]->{startTime};
    $json->{type}        = $bpc_info[$i]->{type};
    $json->{comp_ratio}  = ( $bpc_info[$i]->{sizeNew} > 0 ) ?
        sprintf( "%.2f", 100 - ( $bpc_info[$i]->{sizeNewComp} * 100 / $bpc_info[$i]->{sizeNew} ) )
          :
        0;

    $json->{new_size_avg}    = int $sizes->mean;
    $json->{new_size_median} = int $sizes->median;
    # Some old versions of Statistics::Descriptive (eg, on el5) do not support quantile
    $json->{new_size_q1}     = eval { int $sizes->quantile(1) } || 0;
    $json->{new_size_q3}     = eval { int $sizes->quantile(3) } || 0;
    $json->{enabled}         = ( $conf->{BackupsDisable} > 0 ) ? 0 : 1;
    $json->{total_size}      = $sizes->sum + $json->{full_size} - 2 * $new_size_of_last_full;
    $json->{age}             = time - $bpc_info[$i]->{startTime};

    $json->{max_errors} = $conf->{MaxXferError} || 0;
  }
} elsif ( $entity ) {

  $json = {
    perf  => 0,
    size  => 0,
    hosts => 0,
    bkp   => 0,
    ratio => 0
  };

  my $total_new  = 0;
  my $total_comp = 0;

  foreach my $host ( keys %{ $bpc->HostInfoRead } ) {
    next unless $host =~ m/^(vm_)?\Q$entity\E_.*/;
    my $full_size;

    $json->{hosts}++;

    my $hostConf = $bpc->ConfigDataRead($host);
    my $conf     = { %$mainConf, %$hostConf };
    my $freq     = ( $conf->{FullPeriod} > $conf->{IncrPeriod} ) ? $conf->{IncrPeriod} : $conf->{FullPeriod};
    my $duration = 0;
    my $bkp_num  = 0;
    my $new_size_of_last_full = 0;

    foreach my $backup ( $bpc->BackupInfoRead( $host ) ) {
      next if ( $backup->{type} !~ m/^full|incr$/ );

      # Save the total size of the last full backup
      if ( $backup->{type} eq 'full' ) {
        $full_size             = $backup->{size};
        $new_size_of_last_full = $backup->{sizeNew};
      }

      $json->{size} += $backup->{sizeNew};
      $total_new    += $backup->{sizeNew};
      $total_comp   += $backup->{sizeNewComp};
      $duration     += $backup->{endTime} - $backup->{startTime};
      $bkp_num++;
      $json->{bkp}++;
    }
    # Compute the average cost as the number of hours per day spent
    # to backup this host
    $json->{perf} += ( $bkp_num > 0 ) ? $duration / ( 3600 * $bkp_num * $freq ) : 0;

    # $json->{size} represent the total size used by this host.
    # But we want to substract the new size of the last full, as for this one we
    # do not count sizeNew but size. As we've already added sizeNew we need to substract it 2 times
    $json->{size} += $full_size - 2 * $new_size_of_last_full;
  }
  $json->{ratio} = ( $total_new > 0 ) ? 100 - ( $total_comp * 100 / $total_new ) : 0;

  # Round some values
  foreach my $key ( qw(ratio perf) ) {
    $json->{$key} = sprintf( "%.2f", $json->{$key} );
  }

} else {
  print<<"EOF";

Usage: $0 --host=<host> or --entity=<entity>

EOF
}

# VERBOSE
# print("\n----------------\n");
# print("new_size_median : $json->{new_size_median}");
# print("\n");
# print("new_size_avg : $json->{new_size_avg}");
# print("\n");
# print("new_size_q3 : $json->{new_size_q3}");
# print("\n");
# print("new_size : $json->{new_size}");
# print("\n");
# print("new_size_q1 : $json->{new_size_q1}");
# print("\n");
# print("full_size : $json->{full_size}");
# print("\n----------------\n");
# print("\n----------------\n");

# TOO SMALL ?
my $toosmall = "true";
if ( $json->{new_size} > $json->{new_size_q3} + $json->{new_size_q3} - $json->{new_size_q1} * 1.5 && $json->{new_size} > $json->{new_size_avg} * 6 ) {
  $toosmall = "false";
  }
else {
  $toosmall = "true";
}
# TOO BIG ?
my $toobig = "true";
if ( $json->{new_size} < $json->{new_size_q1} - $json->{new_size_q3} - $json->{new_size_q1} * 1.5 && $json->{new_size} < $json->{new_size_avg} / 3 ) {
  $toobig = "false";
  }
else {
  $toobig = "true";
}

print("Size Consistency : ");
if ( $toosmall == "false" && $toobig == "false" ) {
  print("Normal");
  }
else {
  print("ANOMALOUS");
}
exit(0);