#!/usr/bin/perl -w

# Written by Tim Ellis
#
# use unixToSnmp.pl to get information from a host via SNMP, then
# alert on it accordingly

use strict;
use DBI;
use POSIX qw/setsid/;
use POSIX qw/strftime/;

# variables that need changing
my $unixToSnmp = "/usr/lib/nagios/plugins/unixToSnmp.pl";

# other arguments that may be passed in
my $community = "public";
my $ipAddr = "";
my $posixCommand = "";
my $verbose = "";
my $logFile = "";
my $threshold;
my @extraArgs;

# stuff for the Nagios alert message and alert decision
my $msg = "";
my $warnThresh;
my $critThresh;
my $warnings = 0;
my $criticals = 0;
my $unknowns = 0;

# list of "commands" and the SNMP command they'll execute
my $commandHash;
$commandHash->{'ps'         } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'loadavg'    } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'hostname-s' } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'ifconfig'   } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'route-n'    } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'arp'        } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'df-k'       } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'df-m'       } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";
$commandHash->{'miscinfo'   } =  "$unixToSnmp --community=___community --hostname=___ipAddr --command=___command";

# need at least one arg
if (scalar @ARGV == 0) {
	&doArgs();
	exit 255;
}
# process input arguments
foreach my $arg (@ARGV) {
	if ($arg =~ /^-+h(elp)*$/)          { doArgs(); exit 0; }
	elsif ($arg =~ /^-+hostname=(.+)$/)    { $ipAddr = $1; }
	elsif ($arg =~ /^-+community=(.+)$/)   { $community = $1; }
	elsif ($arg =~ /^-+command=(.+)$/)     { $posixCommand = $1; }
	elsif ($arg =~ /^-+threshold=(.+)$/)   { $threshold = $1; }
	elsif ($arg =~ /^-+verbose$/)          { $verbose = 1; }
	elsif ($arg =~ /^-+logfile=(.+)$/)     { $logFile = $1; }
	else { push (@extraArgs, $arg); }
}

# if they want to log to a file, open it, later doLog() will
# write into this file
if ($logFile) {
	open (LOG, ">> $logFile") || die "Can't open $logFile for writing.";
	close (STDERR);
	open (STDERR, ">> $logFile");
}

# we'll overwrite this from time to time
my $currTime = strftime ('%Y-%m-%d %H:%M:%S', localtime);

# run the command specified
my $cmdToRun = $commandHash->{$posixCommand};
if (!defined $cmdToRun) {
	doArgs();
	exit 255;
}
$cmdToRun =~ s/___community/$community/g;
$cmdToRun =~ s/___ipAddr/$ipAddr/g;
$cmdToRun =~ s/___command/$posixCommand/g;
# tack on all the extra args passed to this script to the callee
foreach my $extraArg (@extraArgs) {
	$cmdToRun .= " $extraArg";
}
my $output = `$cmdToRun`;
chomp ($output);

# now do special processing based on each alert type
if ($posixCommand =~ /^df-[k,m]/) {
	if ($threshold =~ /^([\d,.]+),([\d,.]+)$/) {
		$warnThresh = $1;
		$critThresh = $2;
	} else {
		$msg = "UNKNOWN: Can't parse input threshold argument \"$threshold\"";
		nagiosExit ($msg);
	}
	foreach my $outLine (split (/\n/, $output)) {
		if ($outLine =~ /\s+(.+?)\s+.+pct: (.+?)\s+/) {
			my $pct = $2 + 0;

			if ($pct > $critThresh) {
				$msg .= " [crit: $1 at $pct%]";
				$criticals++;
			} elsif ($pct > $warnThresh) {
				$msg .= " [warn: $1 at $pct%]";
				$warnings++;
			}
		} else {
			$msg .= "bizarre output from subscript: $outLine ; ";
			$unknowns++;
		}
	}
} else {
	print "$output\n";
	$msg = "UNKNOWN: No logic for determining if thresholds passed for command=$posixCommand";
	nagiosExit ($msg);
}

# now decide how to exit the script. Critical? Warning? OK?
if ($criticals > 0) { $msg = "CRITICAL:$msg"; }
elsif ($warnings > 0) { $msg = "WARNING:$msg"; }
elsif ($unknowns > 0) { $msg = "UNKNOWN:$msg"; }

#print "debug: ct:$critThresh/wt:$warnThresh  c:$criticals  w:$warnings  u:$unknowns\n";
if ($msg) { nagiosExit ($msg); } else { nagiosExit ("OK: No thresholds passed."); }

# quit
if ($logFile) { close LOG; }
exit 0;


# -----------------------------------------------------------------
# -----------------------------------------------------------------
#                           subroutines
# -----------------------------------------------------------------
# -----------------------------------------------------------------

# usage
sub doArgs {
	print "\n";
	print "usage: $0 --hostname=<hostName> --command=<command>\n";
	print "       [--community=<community>] [--logfile=<logFileName>] [--verbose] [--help]\n";
	print "  --hostname     host to use to report the stats\n";
	print "  --community    SNMP community to query as\n";
	print "  --command      POSIX command (with nonstandard output) to query\n";
	print "  --threshold    if results exceed these thresholds, return CRITICAL or WARNING, otherwise return OK\n";
	print "  --logfile      send log output to logFileName rather than STDERR\n";
	print "  --help         get help\n";
	print "  --verbose      be verbose\n";
	print "Valid commands for this script are:\n";
	foreach my $key (sort keys %{$commandHash}) {
		printf "  %12s ",$key;
		if ($key =~ /df-[k,m]/)    { print " --threshold=w%,c%   Partition percent full warning/critial\n"; }
		elsif ($key eq 'loadavg')  { print " --threshold=w1,w5,w15,c1,c5,c15** load average warning/critical\n"; }
		elsif ($key eq 'ifconfig') { print " --threshold=w%,c%** Interface percent errors/discards warning/critial\n"; }
		elsif ($key eq 'miscinfo') { print " --threshold=w,W,c,C min,MAX num processes to trigger warning/critical\n"; }
		else { print "\n"; }
	}
	print "  ** This threshold isn't implemented yet.\n";
	print "Examples (for brevity I leave out the --hostname and --community params):\n";
	print "  $0 --command=ifconfig eth0 --threshold=1,2     # (only eth0 interface)\n";
	print "  $0 --command=df   /        --threshold=85,90   # (only / partition)\n";
	print "  $0 --command=df FixedDisk  --threshold=85,90   # (only fixed disks - not, eg, RAM)\n";
	print "  $0 --command=miscinfo      --threshold=25,100,10,1000\n";
	print "  $0 --command=loadavg       --threshold=5,4,3,10,8,6\n";
}

# output something for logging -- if the input is " X yyyyyyy" then stick
# a timestamp in the whitespace between X and yyyy, otherwise, just output
# what we were given.
sub doLog {
	my $toLog = shift;
	my $toPrint;
	$currTime = strftime ('%Y-%m-%d %H:%M:%S', localtime);

	if ($toLog =~ /^\s([^\s])\s(.+)$/) {
		$toPrint = " $1 $currTime :: $2\n";
	} else {
		$toPrint = " - $currTime :: $toLog\n";
	}

	# send output to STDERR or log file, whichever user specified
	if ($logFile) {
		print LOG $toPrint;
	} else {
		print STDERR $toPrint;
	}
}

# run a system command. exit on failure. log the command if $verbose
sub doSystem {
	my $cmd = shift;
	if ($verbose) { &doLog (" + $cmd"); }

	system ($cmd);

	if ($? >> 8) {
		&doLog (" E Child process returned " . $? >> 8 . " exit status. Dying.");
		exit 255;
	}
}

# change single quotes to doubled-up single quotes
sub doQuote {
	my $arg = shift;
	$arg =~ s/'/''/g;
	return $arg;
}

sub nagiosExit {
	my $msg = shift;

	print "$msg\n";

	if ($msg =~ /^OK/) { exit 0; }
	elsif ($msg =~ /^WARNING/) { exit 1; }
	elsif ($msg =~ /^CRITICAL/) { exit 2; }
	elsif ($msg =~ /^UNKNOWN/) { exit 3; }
}