#!/usr/local/bin/perl
##################################################################
# @(#) $Id: mrtg-ping-probe,v 2.4 2002/07/14 19:26:17 pwo Exp $
# @(#) mrtg-ping-probe release $Name: Release_2_1_0 $
#
# Copyright (c) 1997-2002 Peter W. Osel <pwo@pwo.de>.
# All Rights Reserved.
#
# See the file COPYRIGHT in the distribution for the exact terms.
#
##################################################################
#
# "I saw -- from the cathedral -- you were watching me"
#
#	-- Tanita Tikaram, `Cathedral Song'
#	on: Tanita Tikaram, `Ancient Heart', 1988, WEA Records
#
##################################################################
require 5.003;
use Getopt::Std;
use File::Basename;
use Config;

$Prog_name = basename($0);		# Who I am
$Prog_vers = q$Revision: 2.4 $;
$Prog_id = q$Id: mrtg-ping-probe,v 2.4 2002/07/14 19:26:17 pwo Exp $;
$Usage = "Usage: $Prog_name [-hsvV] [-d deadtime] [-k count] [-l length] [-o ping_options] [-p [factor*]{min|max|avg|loss|integer}/[factor*]{min|max|avg|loss|integer}] [-r [rsh:][user@]host[:osname]] [-t timeout] host\n";

# Parse Command Line:
die $Usage unless getopts('d:Dhk:l:o:p:r:st:vV');

# Generic Options
$Debug		= defined($opt_D) ? $opt_D : 0;
$PrintHelp	= defined($opt_h) ? $opt_h : 0;
$Verbose	= defined($opt_v) ? $opt_v : 0;
$Silent		= defined($opt_s) ? $opt_s : 0;
$PrintVersion	= defined($opt_V) ? $opt_V : 0;

# Tool Specific Options
$DeadTime	= defined($opt_d) ? $opt_d : 0;
$PacketCount	= defined($opt_k) ? $opt_k : "10";
$PacketLength	= defined($opt_l) ? $opt_l : "56";
$PingOptions	= defined($opt_o) ? $opt_o : "";
$PickList	= defined($opt_p) ? $opt_p : "max/min";
$RemotePing	= defined($opt_r) ? $opt_r : "";
$TimeOut	= defined($opt_t) ? $opt_t : 0;

# Check Sanity of Arguments
$err += check_numeric("-d", $DeadTime);
$err += check_numeric("-k", $PacketCount);
$err += check_numeric("-l", $PacketLength);
$err += check_numeric("-t", $TimeOut);
$err += check_picklst("-p", $PickList);

if ($err) {
	print STDERR $Usage;
	exit(1);
	}


if ($PrintVersion ) {
	print STDERR "$Prog_name: $Prog_vers\n";
	exit(0);
	}

if ($PrintHelp) {
	print $Usage;
	exit(0);
	}

if (@ARGV > 1) {
	print STDERR "$Prog_name: ERROR: ignoring superfluous arguments\n";
	print STDERR "$Usage";
	}

if (@ARGV < 1) {
	print STDERR "$Prog_name: FATAL: ping what?\n";
	print STDERR "$Usage";
	exit(1);
	}


($HostToPing) = @ARGV;


($pt{min}, $pt{avg}, $pt{max}, $pt{loss}) =
	ping($HostToPing, $PacketLength, $PacketCount, $TimeOut);

print "$Prog_name: DBG: main(): ping() ret: $pt{min}, $pt{avg}, $pt{max}, $pt{loss}\n" if $Debug;

$PickList =~ /^(?:(\d+)\*)?(\w+)\/(?:(\d+)\*)?(\w+)$/;
($f1, $p1, $f2, $p2) = ($1, $2, $3, $4);
$f1 = $f1 ? $f1 : 1;
$f2 = $f2 ? $f2 : 1;
print "$Prog_name: DBG: main(): before psub: f = ($f1 $f2); p = ($p1 $p2)\n" if $Debug;

$p1 = $p1 =~ /^\d+$/ ? $p1 : $pt{$p1};
$p2 = $p2 =~ /^\d+$/ ? $p2 : $pt{$p2};

print "$Prog_name: DBG: main(): after  psub: f = ($f1 $f2); p = ($p1 $p2)\n" if $Debug;

# The external mrtg probe returns up to 4 lines of output:
#	1. Line: current state of the 'incoming bytes counter'
#	2. Line: current state of the 'outgoing bytes counter'
#	3. Line: string, telling the uptime of the target.
#	4. Line: telling the name of the target.
# We leave out line 3, and 4.

printf "%d\n%d\n", $f1 * $p1, $f2 * $p2;

exit(0);


##################################################################
sub check_numeric {
	my($opt, $val) = @_;
	my($err) = 0;

	unless ($val =~ /^\d+$/) {
		print STDERR "$Prog_name: FATAL: option $opt requires numeric argument.\n";
		++$err;
		}
	return($err);
	}

##################################################################
sub check_picklst {
	my($opt, $val) = @_;
	my($err, $i) = (0, 0);
	my(@v, @n);

	unless ($val =~ /^(?:(\d+)\*)?(\w+)\/(?:(\d+)\*)?(\w+)$/) {
		print STDERR "$Prog_name: FATAL: option $opt requires [factor*]word/[factor*]word argument, I found \"$val\"\n";
		++$err;
		}
	@n = ($1, $3);
	@v = ($2, $4);

	print "$Prog_name: DBG: check_picklst(): n = (@n); v = (@v)\n" if $Debug;

	foreach $i (0..1) {
		unless ($v[$i] =~ /^(min|max|avg|loss|\d+)$/) {
			print STDERR "$Prog_name: FATAL: option $opt uses unknown item $v[$i]\n";
			print STDERR "$Prog_name: FATAL: option $opt may choose from min|max|avg|loss or number\n";
			++$err;
			}
		}

	return($err);
	}

##################################################################
# ping selects the external or internal, builtin ping method
# and starts the external ping program optionally with a
# timout to abort external ping programs that seem to hang
# when the target is unreachable.

sub ping {
	my($host, $length, $count, $timeout) = @_;
	my(%pt);

	($pt{min}, $pt{avg}, $pt{max}, $pt{loss}) =
		ext_ping($host, $length, $count, $timeout);

	return($pt{min}, $pt{avg}, $pt{max}, $pt{loss});
	}

##################################################################
# ping host retrieve and return min, avg, max round trip time
# relying on finding a standard ping in PATH.
# Try to not be platform specific if at all possible.
#
sub ext_ping {
	my($host, $length, $count, $timeout) = @_;
	my(%ping, $ping_output, $redirect_stderr, $pid, %pt);
	my($alarm_exists);

	# List of known ping programs
	%ping = (
		'MSWin32'	=> "ping -l $length -n $count $host",
		'aix'		=> "/etc/ping $host $length $count",
		'bsdos'		=> "/bin/ping -s $length -c $count $host",
		'darwin'	=> "/sbin/ping -s $length -c $count $host",
		'dec_osf'	=> "/sbin/ping -s $length -c $count $host",
		'freebsd'	=> "/sbin/ping -s $length -c $count $host",
		'hpux'		=> "/etc/ping $host $length $count",
		'irix'		=> "/usr/etc/ping -s $length -c $count $host",
		'linux'		=> "/bin/ping -s $length -c $count $host",
		'netbsd'	=> "/sbin/ping -s $length -c $count $host",
		'openbsd'	=> "/sbin/ping -s $length -c $count $host",
		'os2'		=> "ping $host $length $count",
		'OS/2'		=> "ping $host $length $count",
		'solaris'	=> "/usr/sbin/ping -s $host $length $count",
		'sunos'		=> "/usr/etc/ping -s $host $length $count",
		);

	unless (defined($ping{$Config{'osname'}})) {
		print STDERR "${Prog_name}: FATAL: Not yet configured for $Config{'osname'}\n";
		exit(1);
		}

	# add ping options, if any
	$ping{$Config{'osname'}} =~ s/ / $PingOptions / if $PingOptions;

	# windows 95/98 does not support stderr redirection...
	# also OS/2 users reported problems with stderr redirection...
	$redirect_stderr = $Config{'osname'} =~ /^(MSWin32|os2|OS\/2)$/i ? "" : "2>&1";

	# freebsd > 3.x does not allow option -s,
	# unless we run as root (which we shouldn't)
	if (($Config{'osname'} =~ /^freebsd$/i) && $>) {
		# remove option -s from ping command
		$ping{$Config{'osname'}} =~ s/ -s \d+//;
		print "$Prog_name: DBG: ext_ping(): ping = ($ping{$Config{'osname'}})\n" if $Debug;
		}

	# initialize return values
	$pt{loss} = 100;
	$pt{min} = $pt{avg} = $pt{max} = $DeadTime;

	# finally call the external ping program and read its output:
	unless ($pid = open(PING, "$ping{$Config{'osname'}} $redirect_stderr |")) {
		print STDERR "${Prog_name}: FATAL: Can't open $ping{$Config{'osname'}}: $!";
		exit(1);
		}

	$alarm_exists = eval { alarm(0); 1 };
	print STDERR "${Prog_name}: WARN: builtin alarm() does not exist, can't timout ping command\n"
		if $Verbose && $timeout && !$alarm_exists;

	if ($alarm_exists) {
		# read and timeout ping() if it takes too long...
		eval {
			local $SIG{ALRM} = sub { die "alarm\n" };	# \n required!
			alarm $timeout;
			while (<PING>) {
				$ping_output .= $_;
				}
			alarm 0;
			};

		if ($@) {
			die unless $@ eq "alarm\n";	# propagate unexpected errors
			# timed out, kill child, get remaining output, ...
			kill $pid;
			while (<PING>) {
				$ping_output .= $_;
				}
			close(PING);

			# ... and set return values to dead values
			unless ($Silent) {
				print STDERR "${Prog_name}: ERROR: external ping hit timeout $timeout, assuming target $host is unreachable\n";
				print STDERR "${Prog_name}: INFO: The output of the ping command $ping{$Config{'osname'}} was:\n";
				print STDERR "$ping_output\n";
				}
			return($pt{min}, $pt{avg}, $pt{max}, $pt{loss});
			}
		}
	else {
		# read and hope that ping() will return in time...
		while (<PING>) {
			$ping_output .= $_;
			}

		}

	# didn't time out, analyse ping output.
	close(PING);

	# try to find round trip times
	if ($ping_output =~ m@(?:round-trip|rtt)(?:\s+\(ms\))?\s+min/avg/max(?:/(?:m|std)-?dev)?\s+=\s+(\d+(?:\.\d+)?)/(\d+(?:\.\d+)?)/(\d+(?:\.\d+)?)@m) {
		$pt{min} = $1; $pt{avg} = $2; $pt{max} = $3;
		}
	elsif ($ping_output =~ m@^\s+\w+\s+=\s+(\d+(?:\.\d+)?)ms,\s+\w+\s+=\s+(\d+(?:\.\d+)?)ms,\s+\w+\s+=\s+(\d+(?:\.\d+)?)ms\s+$@m) {
		# this should catch most windows locales
		$pt{min} = $1; $pt{avg} = $3; $pt{max} = $2;
		}
	else {
		unless ($Silent) {
			print STDERR "${Prog_name}: ERROR: Could not find ping summary for $host\n";
			print STDERR "${Prog_name}: INFO: The output of the ping command $ping{$Config{'osname'}} was:\n";
			print STDERR "$ping_output\n";
			}
		}

	# try to find packet loss
	# ToDo: only if requested?)
	if ($ping_output =~ m@(\d+)% (?:packet )?loss(?:$|,)@m) {
		# Unix
		$pt{loss} = $1;
		}
	elsif ($ping_output =~ m@\(perte\s+(\d+)%\),\s+$@m) {
		# Windows french locale
		$pt{loss} = $1;
		}
	elsif ($ping_output =~ m@\((\d+)%\s+(?:loss|perdidos|de perda|Verlust)\),\s+$@m) {
		# Windows portugesee, spanish, brazilian, german locale
		$pt{loss} = $1;
		}
	else {
		unless ($Silent) {
			print STDERR "${Prog_name}: ERROR: Could not find packet loss summary for $host\n";
			print STDERR "${Prog_name}: INFO: The output of the ping command $ping{$Config{'osname'}} was:\n";
			print STDERR "$ping_output\n";
			}
		}

	return($pt{min}, $pt{avg}, $pt{max}, $pt{loss});
}

__END__

=head1 NAME

mrtg-ping-probe - a round trip time and packet loss probe for MRTG

=head1 SYNOPSIS

B<mrtg-ping-probe>
[ B<-hsvV> ]
[ B<-d> I<deadtime> ]
[ B<-k> I<count> ]
[ B<-l> I<length> ]
[ B<-o> I<ping_options> ]
[ B<-p> [I<factor>*]I<item>/[I<factor>*]I<item> ]
[ B<-r> I<[rsh:][user@]host[:osname]> ]
[ B<-t> I<timeout> ]
I<host>

=head1 DESCRIPTION

B<mrtg-ping-probe> pings the given host I<host> and prints on stdout
two lines extracted from the ping output.  The default is to print
the maximum, and the minimum round trip time.

It is meant to be called by the Multi Router Traffic Grapher (MRTG).


=head1 OPTIONS


=over 8


=item B<-h>

print help on stdout and exit.


=item B<-v>

Be more verbose.


=item B<-V>

Print version number on stderr and exit.


=item B<-d> I<deadtime>

Specifies the value we return for round trip times in case we assume
that the target is down.  The default is zero.  We assume that the
target is unreachable, if we cannot find the ping summary or if the
ping program was aborted because of a set timeout.

For WAN connections that usually have round trip times of 10ms and
higher, ranges of zero round trip time are highly visible.  In a LAN
environment, you might set it to a high value, e.g. 999, which however
might change the scale of the graphs in such a way that you hardly see
the regular round trip times.  You might use mrtg-misc-probe's pong
option to generate a graph that shows reachabilty of targets, instead.


=item B<-k> I<count>

Specifies the number of of ping packets to be sent.  The default is to
send 10 ping packets.


=item B<-l> I<length>

Use I<length> as the length of the data portion of the ICMP ECHO
request packet.  The default I<length> is 56 data bytes.


=item B<-o> I<ping_options>

Pass I<ping_options> to the ping program.  You can use this generic
option to e.g. pass an option to ping to suppress displaying addresses
as host names.  This helps to prevent the ping to fail because it
cannot map hostnames to IP addresses and vice versa.  To pass several
arguments, enclose the options in quotes.  Check the documentation of
your ping program for possible options.


=item B<-p> [I<factor>B<*>]I<item>/[I<factor>B<*>]I<item>

Pick the values you want mrtg-ping-probe to return.  Allowed values for
I<item> are: B<min>, B<max>, B<avg>, B<loss>, or an I<integer>.  Each item
can be preceded by a integer factor used to multiply the value returned
by the ping program.  The default picklist is B<min/max>.

To display ping times in microseconds instead of millisconds, use: B<-p
1000*max/1000*min>.


=item B<-r> I<[rsh:][user@]host[:osname]>

B<Not Yet Implemented>

run ping on remote host I<host>, as user I<user> (or as local user, if
no user is given).  Uses B<rsh -n> to start program on remote host,
unless you provide a different program name.  If the remote host has a
different system type than the local host (if the osname is different)
you have to say so.

This option can be used if you run mrtg on a host that cannot ping to
the final target, and you cannot install mrtg and/or perl on the
intermediate host used to ping the final target.


=item B<-s>

Silent mode.  Do not generate error messages if there is no response
from the ping program or if it ran into the timeout.  Usually cron will
mail you these error messages, which might be helpful to debug
problems.


=item B<-t> I<timeout>

Abort the external ping program after I<timeout> seconds.  A I<timeout>
value of zero (the default) means, we do not abort the external ping
program.

If mrtg-ping-probe seems to hang forever, check your ping program, it
might be a version that wants to B<receive> the given number of
ECHO_RESPONSE packets instead of just sending them.  If your target is
unreachable, these pings ping forever.

You want to choose I<timeout> as short as possible to leave mrtg enough
time for all your other targets, but long enough so you do not abort
pings (too often).  You might use (I<count> * worst case round trip
time) as a starting point.  (Or install a ping program that is not
broken ;-)

If your perl installation does not implement the builtin alarm()
function, the timout option will be ignored.  You will get a warning
about this only in verbose mode (option B<-v>).  I have not found a
perl installation on Windows that implements the alarm() builtin
function on Win32.  So basically on Windows the timout option is
not working.

=back


=head1 RETURN VALUE

The program exits with an exit value 0, if it believes it was
successful.


=head1 EXAMPLES

=over 4

=item B<mrtg-ping-probe ricochet>

Retrieves the maximum and minimum round trip time to the host
B<ricochet>, using the default length and count.


=item B<mrtg-ping-probe -p max/avg ricochet>

Retrieves the maximum and average round trip time to the host
B<ricochet>, using the default length and count.


=item B<mrtg-ping-probe -p '1000*max/1000*avg' ricochet>

Retrieves the maximum and average round trip time to the host
B<ricochet> multiplied by a factor of 1000, using the default length
and count.


=item B<mrtg-ping-probe -k 17 -l 1000 192.168.192.42>

Retrieves the maximum and minimum round trip time to the host
192.168.192.42, using 17 1000 data bytes pings.


=item B<mrtg-ping-probe -o -n ricochet.pwo.de>

Suppress displaying addresses as host names on Solaris 2 (to protect
from DNS problems causing the ping to fail) by passing option B<-n> to
the ping program.

Note that in this example `B<-n>' is not an option for mrtg-ping-probe,
but gets passed to the ping program.


=item B<mrtg-ping-probe -o '-n -I 3' ricochet.pwo.de>

Pass several options B<-n -I 3> to the ping program.


=item B<mrtg-ping-probe -p loss/loss ricochet.pwo.de>

Monitors the packet loss for the link to host ricochet.pwo.de.


=back

=head1 FILES

B<mrtg-ping-probe> uses an external ping program, like
F</usr/sbin/ping>.

=head1 SEE ALSO

mrtg(1), mrtg-ping-cfg, ping(1), mrtg.cfg-ping, mrtg-misc-probe(1)

http://www.mrtg.org/

http://pwo.de/projects/mrtg/

=head1 DIAGNOSTICS

=over 4

=item FATAL: Not yet configured for I<osname>

Currently B<mrtg-ping-probe> depends on an external ping program, which
every operating systems hides in another place.  Also different
programs require different arguments.  We have a configuration table
listing the ping program for each operating systems.  You have to
figure out how to call which program on your platform, and add to the
information to the table.  Please contribute back any additions, so I
can include them in the next version.


=item ERROR: ignoring superfluous arguments

More than one argument was given.  B<mrtg-ping-probe> will ignore all
but the first argument.  The first argument is taken as a hostname or
IP address of an host and B<mrtg-ping-probe> will try to ping it.

=item FATAL: ping what?

No argument was given.  B<mrtg-ping-probe> terminates, as there is
nothing to ping.


=item FATAL: option I<option> requires numeric argument.

The argument for option I<option> was not an integer number.


=item FATAL: Can't open ping: I<some reason>

B<mrtg-ping-probe> was not able to execute the external ping program.
Check the pathname and permissions of the external ping program.
I<some reason> might give some useful hints.


=item ERROR: external ping hit timeout I<timeout>,
assuming target I<host> is unreachable

We ran into a timeout pinging the target host I<host>.  You might have
to increase the timeout value (Option B<-t>) if this happens when the
target is up and the round trip time just happens to be longer than
usual.

The captured output of the ping program is printed and will (hopefully)
give further hints why this problem occurred.

This message is not printed if mrtg-ping-probe runs in silent mode
(Option B<-s>).


=item ERROR: Could not find ping summary for I<host>

B<mrtg-ping-probe> was not able to find the ping summary.  Most likely,
the host is not reachable.  If your operating system changed (e.g. it
was upgraded to a new version, or a new version of the ping program was
installed), it might also be necessary to change the regular expression
that extracts the round trip times.  You might want to use the perl
script check-ping-fmt (which is part of the soure distribution) to test
the regular expression.

The captured output of the ping program is printed and will (hopefully)
give further hints why this problem occurred.

This message is not printed if mrtg-ping-probe runs in silent mode
(Option B<-s>).



=item ERROR: Could not find packet loss summary for I<host>

B<mrtg-ping-probe> was not able to find the packet loss summary.
If your operating system changed (e.g. it was upgraded to a new
version, or a new version of the ping program was installed), it might
also be necessary to change the regular expression that extracts the
packet loss.  You might want to use the perl script check-ping-fmt
(which is part of the soure distribution) to test the regular
expression.

The captured output of the ping program is printed and will (hopefully)
give further hints why this problem occurred.

This message is not printed if mrtg-ping-probe runs in silent mode
(Option B<-s>).


=back

=head1 RESTRICTIONS

B<mrtg-ping-probe> currently depends on an external ping(1) program.
If the external program does not support an option, the option given to
B<mrtg-ping-probe> will be ignored.

Under B<freebsd> release 3.x or later, ping option B<-s>
I<packet-length> (B<mrtg-ping-probe> option B<-l> I<length>) is only
allowed to be used when we run as root (which we should not), therefore
this option is silently removed on B<freebsd> before we call the
external ping program.

=head1 BUGS

This program has way too many options and tries to support too many
different systems.

Using this program to monitor sub-millisecond round trip times or
packet loss might be questionable.

Option B<-r>, remote execution of the pirn program, is not yet
implemented.


=head1 COPYRIGHT

Copyright (c) 1997-2002 Peter W. Osel <pwo@pwo.de>.
All Rights Reserved.

See the file COPYRIGHT in the distribution for the exact terms.

=head1 AUTHOR

Written by Peter W. Osel E<lt>pwo@pwo.deE<gt>.
http://pwo.de/

