Issues with network connectivity and Net::Ping

[Steve_BZ]

Hi Guys,

I have an active wireless connection on my Linux box (Kubuntu 13.10). When I manually ping Google with ping www.google.com at the command prompt I get the following output:

PING www.google.com (173.194.34.144) 56(84) bytes of data.
64 bytes from lhr14s21-in-f16.1e100.net (173.194.34.144): icmp_seq=1 t
+tl=57 time=159 ms
64 bytes from lhr14s21-in-f16.1e100.net (173.194.34.144): icmp_seq=2 t
+tl=57 time=190 ms
64 bytes from lhr14s21-in-f16.1e100.net (173.194.34.144): icmp_seq=3 t
+tl=57 time=72.6 ms
64 bytes from lhr14s21-in-f16.1e100.net (173.194.34.144): icmp_seq=4 t
+tl=57 time=199 ms
64 bytes from lhr14s21-in-f16.1e100.net (173.194.34.144): icmp_seq=5 t
+tl=57 time=69.4 ms

But when I use Net::Ping in a perl script like this:

    my $p = Net::Ping->new("tcp", 2);
    # With or without this next line.
    $p->port_number(getservbyname("http", "tcp"));
    if ($p->ping("www.google.com")) {
        print "Host is reachable\n";
        return 1;
    } else {
        print "Host is not reachable\n";
        return 0;
    }

I get "Host is not reachable" rather than the "Host is reachable" suggested by running ping manually.

What is happening here?

Thanks very much.

Steve.

[VinsWorldcom]

I think your problem is 'getservbyname' is not returning what you think it is. This works fine for me (Win7 x64 / Strawberry 5.18.1):

VinsWorldcom@C:\Users\VinsWorldcom\tmp> cat test.pl
use strict;
use warnings;
use Net::Ping;

my $p = Net::Ping->new("tcp", 2);

printf "GetServByName = %s\n", getservbyname("http", "tcp");

# Explicitly use the HTTP port number
$p->port_number(80);
if ($p->ping("www.google.com")) {
    print "Host is reachable\n";
} else {
    print "Host is not reachable\n";
}

VinsWorldcom@C:\Users\VinsWorldcom\tmp> test.pl
GetServByName = http
Host is reachable

[Steve_BZ]

It's true. I just tested it and it works on Linux too.

I'll time them both and see if I can get below the 500 ms that upset Karl so much. I think we should try to eliminate the interference of Solar Flares.

[Steve_BZ]

Hi Guys,

Well here are the timings (the code is at the bottom for you to critique):

It is using a mobile router from three.com, so it may be less stable than a fibre optic connection.

1) Here is a straight ping at t the command prompt (Range 75-400 ms):

PING www.google.com (173.194.34.146) 56(84) bytes of data.
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=1 t
+tl=55 time=394 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=2 t
+tl=55 time=332 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=3 t
+tl=55 time=362 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=4 t
+tl=55 time=276 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=5 t
+tl=55 time=92.4 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=6 t
+tl=55 time=219 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=7 t
+tl=55 time=116 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=8 t
+tl=55 time=265 ms
64 bytes from lhr14s21-in-f18.1e100.net (173.194.34.146): icmp_seq=9 t
+tl=55 time=75.9 ms
^C

2) Here is the data from the programmatic pings:

using Net::Ping  (Range 500-1000 ms)
1031.06737136841 ms
729.610681533813 ms
1004.48346138 ms
775.990009307861 ms
600.036859512329 ms
589.807271957397 ms
754.30154800415 ms
605.571508407593 ms
560.036897659302 ms
782.053709030151 ms
using Net::Ping::External (Range 400-800 ms)
791.388988494873 ms
570.100784301758 ms
687.773942947388 ms
402.096509933472 ms
430.928707122803 ms
398.749589920044 ms
712.586641311646 ms
837.354421615601 ms
560.385465621948 ms
650.154113769531 ms

3) Here is the program:

#!/usr/bin/perl -w
use strict;
use Time::HiRes qw( usleep ualarm gettimeofday tv_interval nanosleep
                    clock_gettime clock_getres clock_nanosleep clock
                    stat lstat );
use strict;
use warnings;
use Net::Ping;


print "using Net::Ping\n";
my $p = Net::Ping->new("tcp", 2);

$p->port_number(80);
my $basetime = clock_gettime();
my $realtime = clock_gettime();
my $roundtrip = ($realtime-$basetime) * 1000;
for (my $i=1; $i<=10 ; $i++){
    if ($p->ping("www.google.com")) {
        $realtime = clock_gettime();
        $roundtrip = ($realtime-$basetime) * 1000;
        print  $roundtrip," ms\n";
    } else {
        print "Host is not reachable\n";
    }
    $basetime = clock_gettime();
}


use Net::Ping::External qw(ping);

print "using Net::Ping::External\n";
#my $p = Net::Ping->new("tcp", 2);

#$p->port_number(80);
    $basetime = clock_gettime();
    $realtime = clock_gettime();
    $roundtrip = ($realtime-$basetime) * 1000;
    for (my $i=1; $i<=10 ; $i++){
        if (ping(host => "www.google.com")) {
            $realtime = clock_gettime();
            $roundtrip = ($realtime-$basetime) * 1000;
            print  $roundtrip," ms\n";
        } else {
            print "Host is not reachable\n";
        }
        $basetime = clock_gettime();
    }

I'm interested in your opinions.

Actually everything takes too long. Rather than doing it synchronously, I'm thinking of setting a global variable which I set from a background loop every ten seconds. That way the information is recent, but i don't need to wait for it.

Then I'll check the variable not the ping.

Regards

Steve

[vinoth.ree]

Could you please try with different protocol instead of 'tcp'. like try with 'icmp' or 'udp' and check.

[FloydATC]

If I recall correctly, the ICMP protocol requires root access, this could be the reason he's trying to use TCP.

I ended up using "/bin/ping" for my own network monitoring script and I believe this was one of the reasons. I also recall having problems with Net::Ping not timing out properly in some situations which can be a real problem when you have more than a handful of hosts to check.

[Steve_BZ]

Hi vinoth.ree,

I tried with udp and I still get a timeout.

FloydATC is correct in supposing that I didn't use ICMP to avoid the need to be root.

I'm thinking of going down the direct call route. What do you think?

Like this

my $p = `ping -c 1 www.google.com | grep -i "bytes.*ms"`

And then I either get:

64 bytes from lhr14s21-in-f16.1e100.net (173.194.34.144): icmp_seq=1 t
+tl=55 time=549 ms

or ..

ping: unknown host www.google.com

There is a slight delay. But it seems reliable. Any views?

Regards

Steve

[Corion]

Potentially, Net::Ping::External, together with the 'external' protocol should accomplish the same.

[karlgoethebier]

IMHO 549ms isn't slight.

Regards, Karl