cqr-cryeye-forks / dns-parallel-prober

PoC for an adaptive parallelised DNS prober

DNS Queue - A Parallelised DNS Prober

What is DNS Parallel Prober?

This is a parallelised domain name prober to find as many subdomains of a given domain as fast as possible.

PLEASE NOTE this script probes DNS servers actively, so please use at your own risk. You are likely to get blacklisted and/or saturate your bandwidth. Whatever you do, it's your responsibility to make sure you have approval for it.

Hat tip to: Kyle F. for the original idea and to ZephrFish for all improvements and testing.

Quickstart

1. Install the requirements:

   pip install -r requirements.txt
   # if you don't want to install stuff as root, do:
   # virtualenv venv
   # source venv/bin/activate
   # pip install -r requirements.txt
   

2. Scan all subdomains of example.com using the wordlist subdomains.txt, using 100 threads. Save the results in out.txt:

   ./dns-queue.py example.com 100 out.txt -i subdomains.txt --simulate
   

   Remove the --simulate part to really scan it.

3. Example run:

    $ python3 dns-queue.py example.com 100 out.txt -i subdomains-short.txt --simulate -f -e err.txt
    [*] SIMULATION IN PROGRESS
    [+] Output destination: 'out.txt'
    [+] Output destination will be overwritten.
    [+] Press CTRL-C to gracefully stop...
    [+] Finding authoritative name servers for domain...
    [+] Using name servers: ['199.43.135.53', '199.43.133.53']
    [+] Checking wildcard DNS...
    [+] Will search for subdomains contained in
    'subdomains-short.txt'
    [+] Saving results to out.txt...
    [+] DNS probing starting...
    100% (200 of 200)
    |##################################################| Elapsed
    Time: 0:00:01 Time: 0:00:01
   
    [+] DNS probing done.
    [+] Waiting for all threads to finish...
    [+] Done.
   

Please note: the --simulate flag will return random results. This is by design.

Other features

If you want to bruteforce all subdomains (default length: 3), use:

./dns-queue.py example.com 100 out.txt

If you have a very fast upstream and don't mind flooding DNS servers, use 2000 threads:

./dns-queue.py example.com 2000 out.txt

By default the script uses the authoritative NS servers for the given domain. To use different DNS servers:

./dns-queue.py example.com 100 out.txt -i subdomains.txt -n ns1.example.com, -n ns2.example.com

For help and other options (e.g. subdomain length, DNS timeouts, etc.):

./dns-queue.py -h

To stop: press ctrl-c - it will wait for the last threads to finish before exiting.

FAQs

Why 100 threads?

The optimal number of threads depends on:

• How fast is the uplink
• How quickly the domain name servers respond to queries

The value is best set empirically: on the same server, try with a value of 100 and keep doubling it until it doesn't go any faster.

In more details, the script creates N threads (here N is 100) and gives each thread a domain name to resolve. It then sleeps a small amount of time before checking whether each thread is done or not. The 'sleep time' is adjusted depending on how fast the threads resolve.

Since network I/O is orders of magnitude slower than CPU I/O, the number of threads should not be limited by the number of cores. In other words: each thread will spend most of its time waiting for a DNS response; that "idle time" can safely be used by other threads.

How do I install pip and virtualenv?

git clone .....
cd dns-parallel-prober
# if debian/ubuntu:
sudo apt-get install python-virtualenv python-pip
# create virtualenv to install the required python libs
virtualenv venv
# activate it
source venv/bin/activate
pip install dnspython
# to deactivate the virtualenv run:
# deactivate

What if I want to use more cores / Why not using Multiprocess?

If you have lots of cores and you can send out data faster than your CPU can fork threads and you want to max out your machine then the simplest solution is:

1. Split a subdomain list into N files (with N matching the number of cores you want to use)
2. Run this executable N times perhaps using tmux, feeding it each file
3. Make sure you use a different output for each process
4. ...
5. Profit!

Alternatively, fork this repo and write a multiprocessing version. Good luck.

Why threads and not processes?

Because in this scenario the bottleneck is the network, not the CPU. I'm happy to be proven wrong! Just fork this repo and submit a pull request and some empirical data to back your claim.

Demo Video

This is a demo of an older version:

Notes

The key thing is that the iteration frequency is dynamically adapted to the depletion speed, i.e. the faster the threads complete the sooner new ones will be added until an equilibrium is reached. The tool tries to stay reasonably close to the maximum value set.

File subdomains.txt gathered from research carried out in 2014/15.