Awesome
What
This repository contains the Python bindings for Dublin Traceroute . To install these bindings you have first to install Dublin Traceroute.
How
To install
You need the following dependencies:
- libpcap
- libtins
- jsoncpp
- libdublintraceroute
You can optionally install pandas
and matplotlib
for additional analysis functionality.
See the following sections for system-specific instructions.
Prerequisites on Debian
libtins
is available in thetesting
repolibdublintraceroute
is available in theunstable
repo
Once you have enabled those repos, install the dependencies as root:
apt-get install libpcap-dev libjsoncpp-dev libtins-dev libdublintraceroute-dev
then run
pip3 install -r requirements.txt
replace pip3
with pip
or pip2
if you are still using python 2. note that
python 2 support will be removed in the future.
Prerequisites on OS X
libtins
andjsoncpp
are available in brewlibpcap
is already installed on OS Xlibdublintraceroute
would like to be in homebrew, but their maintainers say that we don't have enough stars on GitHub to accept the formula. Hence it can be installed from source, or using the formula I wrote. See below or see https://github.com/insomniacslk/dublin-traceroute/blob/master/documentation/readme/README.md#building-on-os-x.
brew install jsoncpp libtins
https://github.com/insomniacslk/dublin-traceroute/blob/master/homebrew/dublin-traceroute.rb
If you prefer to install libdublintraceroute
from source instead, see https://dublin-traceroute.net .
Then run
pip3 install -r requirements_osx.txt
replace pip3
with pip
or pip2
if you are still using python 2. note that
python 2 support will be removed in the future.
If the installation fails on pygraphviz
, try one of the two following methods:
Method 1:
export CFLAGS="-I $(brew --prefix graphviz)/include"
export LDFLAGS="-L $(brew --prefix graphviz)/lib"
pip install -r requirements_osx.txt
Method 2:
pip install \
--global-option=build_ext \
--global-option="-I$(brew --prefix graphviz)/include/" \
--global-option="-L$(brew --prefix graphviz)/lib/" \
pygraphviz
Installing dublintraceroute
Once the prerequisites are installed, there are two ways to install dublintraceroute
itself. From source, or from PyPI (i.e. using pip
or easy_install
)
From source
python setup.py install
If you do not want to install the module as root, just add --user
to the setup.py invocation.
From PyPI
pip install dublintraceroute # use pip3 for the Python3 version
To run
Let's try to run the same traceroute in the CLI example, but this time using Python. Remember that you need root permissions in this case, or you need to manually set the CAP_NET_RAW capability to your Python interpreter (which is not necessarily a good idea).
Let's use 12345 as source port, 33434 as base destination port, and 8.8.8.8 as target:
>>> import dublintraceroute
>>> dublin = dublintraceroute.DublinTraceroute('8.8.8.8', npaths=3)
>>> print(dublin)
<DublinTraceroute (target='8.8.8.8', sport=12345, dport=33434, npaths=3, max_ttl=30)>
>>> results = dublin.traceroute()
>>> type(results)
<class 'dublintraceroute.tracerouteresults.TracerouteResults'>
>>> pprint.pprint(results)
{u'flows': {u'33434': [{u'is_last': False,
u'name': u'gateway',
u'nat_id': 0,
u'received': {u'icmp': {u'code': 11,
u'description': u'TTL expired in transit',
u'extensions': [],
u'mpls_labels': [],
u'type': 0},
u'ip': {u'dst': u'192.168.9.20',
u'id': 1999,
u'src': u'192.168.9.1',
u'ttl': 64},
u'timestamp': u'1474069492.648438'},
u'rtt_usec': 308179,
u'sent': {u'ip': {u'dst': u'8.8.8.8',
u'src': u'192.168.9.20',
u'ttl': 1},
u'timestamp': u'1474069492.340259',
u'udp': {u'dport': 33434,
u'sport': 12345}}},
...
>>> results.pretty_print()
ttl 33436 33434 33435
----- -------------------------------------------- -------------------------------------------- --------------------------------------------
1 gateway (427110 usec) gateway (308179 usec) gateway (403034 usec)
2 * * *
3 188-141-126-1.dynamic.upc.ie (433980 usec) 188-141-126-1.dynamic.upc.ie (355249 usec) 188-141-126-1.dynamic.upc.ie (414672 usec)
4 ie-dub01a-ra3-ae34-0.aorta.net (448009 usec) ie-dub01a-ra3-ae34-0.aorta.net (141946 usec) ie-dub01a-ra3-ae34-0.aorta.net (414890 usec)
5 ie-dub01a-ri1-ae50-0.aorta.net (435168 usec) 84-116-130-142.aorta.net (355207 usec) 84-116-130-142.aorta.net (419248 usec)
6 213.46.165.18 (460317 usec) 213.46.165.18 (355184 usec) 213.46.165.18 (420044 usec)
7 216.239.43.3 (460323 usec) 216.239.43.3 (355110 usec) 66.249.95.91 (420293 usec)
8 * google-public-dns-a.google.com (355082 usec) google-public-dns-a.google.com (419689 usec)
9 * google-public-dns-a.google.com (355036 usec) google-public-dns-a.google.com (419759 usec)
10 * google-public-dns-a.google.com (354993 usec) google-public-dns-a.google.com (419943 usec)
11 * google-public-dns-a.google.com (354945 usec) google-public-dns-a.google.com (421734 usec)
12 * google-public-dns-a.google.com (381269 usec) google-public-dns-a.google.com (423064 usec)
13 * google-public-dns-a.google.com (381319 usec) google-public-dns-a.google.com (423643 usec)
14 * * *
15 * * *
16 * * *
17 * * *
18 * * *
19 * * *
20 * * *
21 * * *
22 * * *
23 * * *
24 * * *
25 * * *
26 * * *
27 * * *
28 * * *
29 * * *
30 * * *
>>> graph = dublintraceroute.to_graphviz(results)
>>> graph.draw('traceroute.png')
>>> graph.write('traceroute.dot')
A naive implementation of the traceroute with a oneliner could be:
$ sudo python -c "import dublintraceroute; dublintraceroute.DublinTraceroute('8.8.8.8', npaths=3).traceroute().pretty_print()"
ttl 33436 33434 33435
----- --------------------------------------------- -------------------------------------------- --------------------------------------------
1 gateway (568664 usec) gateway (357115 usec) gateway (391580 usec)
2 * * *
3 188-141-126-1.dynamic.upc.ie (639608 usec) 188-141-126-1.dynamic.upc.ie (357118 usec) 188-141-126-1.dynamic.upc.ie (458835 usec)
4 ie-dub01a-ra3-ae34-0.aorta.net (718752 usec) ie-dub01a-ra3-ae34-0.aorta.net (69098 usec) ie-dub01a-ra3-ae34-0.aorta.net (416961 usec)
5 ie-dub01a-ri1-ae50-0.aorta.net (846613 usec) 84-116-130-142.aorta.net (357079 usec) 84-116-130-142.aorta.net (470277 usec)
6 213.46.165.18 (867147 usec) 213.46.165.18 (363147 usec) 213.46.165.18 (470332 usec)
7 216.239.43.3 (867160 usec) 216.239.43.3 (363183 usec) 66.249.95.91 (470292 usec)
8 * google-public-dns-a.google.com (363209 usec) google-public-dns-a.google.com (470243 usec)
9 * google-public-dns-a.google.com (363214 usec) google-public-dns-a.google.com (470228 usec)
10 * google-public-dns-a.google.com (363211 usec) google-public-dns-a.google.com (470172 usec)
11 * google-public-dns-a.google.com (363204 usec) google-public-dns-a.google.com (470146 usec)
12 * google-public-dns-a.google.com (363197 usec) google-public-dns-a.google.com (470125 usec)
13 * google-public-dns-a.google.com (363190 usec) google-public-dns-a.google.com (475719 usec)
14 * * *
15 * * *
16 * * *
17 * * *
18 * * *
19 * * *
20 * * *
21 google-public-dns-a.google.com (1359710 usec) * *
22 * * *
23 * * *
24 * * *
25 * * *
26 * * *
27 * * *
28 * * *
29 * * *
30 * * *
You can also invoke the module directly, with python -m dublintraceroute --help
.
For example:
$ sudo python -m dublintraceroute trace 8.8.8.8
...
then generate a PNG from the traceroute:
python -m dublintraceroute plot trace.json
Sending probes
You can also send simple probes, without running a full traceroute. This
translates in sending only one packet per path, at a higher TTL. This
functionality is wrapped in the probe
command:
$ sudo python3 -m dublintraceroute probe google.com
Sending probes to google.com
Source port: 12345, destination port: 33434, num paths: 20, TTL: 64, delay: 10, broken NAT: False
# target src port dst port rtt (usec)
--- ------------- ---------- ---------- ------------
1 216.58.198.78 12345 33434 15705
2 216.58.198.78 12345 33435 15902
3 216.58.198.78 12345 33436 16127
4 216.58.198.78 12345 33437 16512
5 216.58.198.78 12345 33438 16465
6 216.58.198.78 12345 33439 11271
7 216.58.198.78 12345 33440 21903
8 216.58.198.78 12345 33441 16811
9 216.58.198.78 12345 33442 16863
10 216.58.198.78 12345 33443 16884
11 216.58.198.78 12345 33444 17225
12 216.58.198.78 12345 33445 12127
13 216.58.198.78 12345 33446 12274
14 216.58.198.78 12345 33447 17783
15 216.58.198.78 12345 33448 17933
16 216.58.198.78 12345 33449 12659
17 216.58.198.78 12345 33450 12749
18 216.58.198.78 12345 33451 18041
19 216.58.198.78 12345 33452 12762
20 216.58.198.78 12345 33453 18404
Data analysis
Pandas DataFrame
python-dublin-traceroute
also supports pandas for data analysis. You can easily analyze a traceroute after converting it to a pandas.DataFrame
:
>>> import dublintraceroute
>>> df = dublintraceroute.DublinTraceroute('8.8.8.8', npaths=3).traceroute().to_dataframe()
>>> import pandas
>>> pandas.set_option('display.width', 180)
>>> df.head()
is_last name nat_id received_icmp_code received_icmp_description received_icmp_extensions \
0 False gateway 0 11 TTL expired in transit []
1 False NaN NaN NaN NaN
2 False 188-141-126-1.dynamic.upc.ie 20530 11 TTL expired in transit []
3 False ie-dub01a-ra3-ae34-0.aorta.net 20530 11 TTL expired in transit [{u'size': 8, u'type': 1, u'class': 1, u'paylo...
4 False ie-dub01a-ri1-ae50-0.aorta.net 20530 11 TTL expired in transit []
received_icmp_mpls_labels received_icmp_type received_ip_dst received_ip_id received_ip_src received_ip_ttl received_timestamp rtt_usec \
0 [] 0 192.168.9.20 1999 192.168.9.1 64 1474070132.380299 1126867
1 NaN NaN NaN NaN NaN NaN NaN NaN
2 [] 0 192.168.9.20 0 188.141.126.1 253 1474070132.380312 1126811
3 [{u'ttl': 1, u'bottom_of_stack': 1, u'experime... 0 192.168.9.20 8405 84.116.238.50 252 1474070132.378559 1125021
4 [] 0 192.168.9.20 0 84.116.130.97 251 1474070132.380326 1126755
sent_ip_dst sent_ip_src sent_ip_ttl sent_timestamp sent_udp_dport sent_udp_sport
0 8.8.8.8 192.168.9.20 1 1474070131.253432 33436 12345
1 8.8.8.8 192.168.9.20 2 1474070131.253468 33436 12345
2 8.8.8.8 192.168.9.20 3 1474070131.253501 33436 12345
3 8.8.8.8 192.168.9.20 4 1474070131.253538 33436 12345
4 8.8.8.8 192.168.9.20 5 1474070131.253571 33436 12345
RTT chart per path
If you have matplotlib
installed, you can also create diagrams of various types. For example, let's visualize the RTT hop-by-hop for each network path:
>>> import matplotlib.pyplot as plt
>>> import dublintraceroute
>>> df = dublintraceroute.DublinTraceroute('8.8.8.8').traceroute().to_dataframe()
>>> group = df.groupby('sent_udp_dport')['rtt_usec']
>>> fig, ax = plt.subplots(figsize=(10, 6))
>>> for label, sdf in group:
>>> sdf.reset_index().plot(ax=ax, label=label, legend=True)
>>> ax.set_title('RTT per destination port')
>>> ax.legend(
>>> [dport for dport, _ in group],
>>> title='destination port',
>>> loc='upper left',
>>> )
>>> plt.show()
>>> fig.savefig('rtt.png')
You will see something like this:
This diagram highlights a latency bump around hops with TTL 3 to 5, and devices not responding at TTL 9 to 12.
Who
Andrea Barberio, find more about me at https://insomniac.slackware.it