Subject: RE: [Bacula-users] monitoring bacula with Nagios
From: "Julian Hein" <jhein@netways.de>
To: <bacula-users@lists.sourceforge.net>

Hi,

> Anyway: I would really like to write such a check_bacula 
> plugin. I just 
> don't know what I need to implement to achive a successful 
> authentication. And maybe to get some infos out. Like current 
> number of 
> jobs, runtime or so.

We are checking bacula with Nagios in two ways: First we check all servers if the neccessary services are running, like the fd on all bacula clients (windows & linux), directors, sd, etc. And the second check is to look in baculas mysql database if there is a successful job for every host within the last 24 hours:


1. Check if the fd is running
=============================

Services:
---------

# bacula-fd linux
check_command                   check_spezial_procs_by_ssh!2:!1:!bacula-fd

# bacula-sd
check_command                   check_spezial_procs_by_ssh!2:!1:!bacula-sd

# bacula-dir
check_command                   check_spezial_procs_by_ssh!2:!1:!bacula-dir

# bacula-fd windows
check_command                   check_nt_service!bacula

Commands:
---------

# check for services by name with ssh
define command {
  command_name  check_spezial_procs_by_ssh
  command_line  $USER1$/check_by_ssh -t 60 -H $HOSTADDRESS$ -C "/opt/nagios/libexec/check_procs -w $ARG1$ -c $ARG2$ -C $ARG3$"
}

# check for the bacula-fd on windows with nsclient
define command {
  command_name  check_nt_service
  command_line  $USER1$/check_nt -H $HOSTADDRESS$ -p portno. -s password -v SERVICESTATE -l $ARG1$
}

2. Is there a successful job in the database
============================================

Services:
---------

# bacula jobs
check_command    check_bacula_by_ssh!27!1!1

Commands:
---------
The name of our backup jobs have to match the hostname in Nagios. So we can check on the backup server, for a job called $HOSTNAME$:

define command {
  command_name    check_bacula_by_ssh
  command_line    $USER1$/check_by_ssh  -t 60 -H my.backup.server -C "/opt/nagios/libexec/check_bacula.pl -H $ARG1$ -w $
ARG2$ -c $ARG3$ -j $HOSTNAME$"
}

check_bacula.pl:
----------------

#!/usr/bin/perl -w
use strict;
use POSIX;
use File::Basename;
use DBI;
use Getopt::Long;
use vars qw(
       $opt_help
           $opt_job
           $opt_critical
           $opt_warning
           $opt_hours
           $opt_usage
           $opt_version
           $out
           $sql
           $date_start
           $date_stop
           $state
           $count
           );
           
sub print_help();
sub print_usage();
sub get_now();
sub get_date;

my $progname = basename($0);

my %ERRORS = (  'UNKNOWN'       =>      '-1',
                'OK'            =>      '0',
                'WARNING'       =>      '1',
                'CRITICAL'      =>      '2');

Getopt::Long::Configure('bundling');
GetOptions
        (
        "c=s"   =>      \$opt_critical, "critical=s"    =>      \$opt_critical,
        "w=s"   =>      \$opt_warning,  "warning=s"     =>      \$opt_warning,
        "H=s"   =>      \$opt_hours,    "hours=s"       =>      \$opt_hours,
        "j=s"   =>      \$opt_job,      "job=s"         =>      \$opt_job,
        "h"     =>      \$opt_help,     "help"          =>      \$opt_help,
                                        "usage"         =>      \$opt_usage,
        "V"     =>      \$opt_version,  "version"       =>      \$opt_version
        ) || die "Try '$progname --help' for more information.\n";

sub print_help() {
print "\n";
print "PRINT HELP...\n";
print "\n";
}

sub print_usage() {
 print "PRINT USAGE...\n";
 print "\n";
}

sub get_now() {
 my $now  = defined $_[0] ? $_[0] : time;
 my $out = strftime("%Y-%m-%d %X", localtime($now));
 return($out);
}

sub get_date {
 my $day = shift;
 my $now  = defined $_[0] ? $_[0] : time;
 my $new = $now - ((60*60*1) * $day);
 my $out = strftime("%Y-%m-%d %X", localtime($new));
 return ($out);
}

if ($opt_help) {
 print_help();
 exit $ERRORS{'UNKNOWN'};
}

if ($opt_usage) {
 print_usage();
 exit $ERRORS{'UNKNOWN'};
}

if ($opt_version) {
 print "$progname 0.0.1\n";
 exit $ERRORS{'UNKNOWN'};
}


if ($opt_job && $opt_warning && $opt_critical) {
 my $dsn = "DBI:mysql:database=bacula;host=localhost";
 my $dbh = DBI->connect( $dsn,'root','' ) or die "Error connecting to: '$dsn': $DBI::errstr\n";
 
 if ($opt_hours)
 {
  $date_stop = get_date($opt_hours);
 }
  else
  {
   $date_stop = '1970-01-01 01:00:00';
  }
 
 $date_start = get_now();
 
 $sql = "SELECT count(*) as 'count' from Job where (Name='$opt_job') and (JobStatus='T') and (EndTime <> '') and ((EndTime <= '$date_start') and (EndTime >= '$date_stop'));";

 my $sth = $dbh->prepare($sql) or die "Error preparing statemment",$dbh->errstr;
 $sth->execute;
 
 while (my @row = $sth->fetchrow_array()) {
  ($count) = @row;
 }
$state = 'OK';
if ($count<$opt_warning) { $state='WARNING' }
if ($count<$opt_critical) { $state='CRITICAL' }

print "Bacula $state: Found $count successfull jobs\n";
exit $ERRORS{$state};
 $dbh->disconnect();
}
 else {
  print_usage();
 }

Well, this script is not really finished, but it works for us. Maybe it is helpful for you. If somebody makes enhancements, I would be happy to recieve a copy.

cu,
Julian

-- 
Julian Hein                   NETWAYS GmbH
Managing Director             Deutschherrnstr. 47a
Fon.0911/92885-0              D-90429 Nürnberg
Fax.0911/92885-31                                        
jhein@netways.de              www.netways.de