Password Cracking on Beoshock with Hashcat and John

Getting on Beoshock

Applying

https://wichita.teamdynamix.com/TDClient/1907/Portal/Requests/ServiceDet?ID=41427
List your intended use as something like “NCL Password Cracking” and your academic advisor as Sergio’s email (omitted for web).

Connecting

You’ll ssh into the head node at first, this is where most of the prep happens.
Use ssh [email protected] with your actual student ID. This should work anywhere on-campus, it’s whitelisted.
When it’s time to crack, use an srun or sbatch to use the actually powerful nodes (we’ll cover that in a following section).

Building Hashcat

Pretty simple, load GCC, and run the makefile. This should include support for CUDA, even though we haven’t loaded the module yet.

1
2
3
4
5
6


module load GCC

git clone https://github.com/hashcat/hashcat.git
cd hashcat
make
./hashcat --version

Building John

Same idea, load GCC, run make.

1
2
3
4
5
6


module load GCC

git clone https://github.com/openwall/john.git
cd john
cd src
./configure && make

Then to run from your home directory (assuming you cloned it there), do ./john/run/john.

Using Slurm

Interactive

Use the following command to spawn an interactive shell on a node with a GPU.

srun -c1 --mem-per-cpu=8G --gres=gpu:1 --pty bash -i

-c1 Request a single CPU
--mem-per-cpu=8G 8GB of RAM (I think total, not per CPU?)
--gres=gpu:1 Request a GPU. Can go up to 4 I believe.
--pty bash Request a pseudo-TTY and use bash
-i Interactive

You may be tempted to increase the resources here, please remember this is a shared resource.
Also remember that a single Beoshock GPU is already wicked powerful.

Queued job

Create an .sh file, we’ll call ours cracking.sh.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11


#!/bin/bash
#SBATCH --job-name=NCL-password-cracking
#SBATCH --time=1:00:00 # your program will be terminated after this
#SBATCH --ntasks=2 # number of processor cores (i.e. tasks)
#SBATCH --nodes=1 # number of nodes
#SBATCH --mem-per-cpu=8G # memory (it says per CPU but I think it's total)
#SBATCH --gres=gpu:1 # request 1 gpu

module load CUDA

~/hashcat/hashcat -m 0 -a 0 hash_file.txt dictionary.txt -r your_rule.rule -o hash_file_results.txt

Now you can queue it to be run with sbatch cracking.sh.
This will run it whether you’re logged in or not, e.g. if there aren’t enough resources available to run your request at the moment.
Want to check if it’s done? Use squeue -u A123B456 with your student ID to see your queued and running jobs.
Want to check how long it took? Use sacct -l -j <job_id> | less. It’s wide, so use less. You can ge the job id from the log file, or if you have it from the sbatch or squeue.
A file named like slurm-786012.out will also be created, with the STDOUT from your program, in case you forgot that -o on hashcat.

Loading files on and off of Beoshock

The easy way (SSH and copy paste)

While SSH’d in, edit some file, and paste into it (shift+ins or right click often).
Handy for most early hashes where they give you a plain hash.

Via OnDemand

If you login to OnDemand here: https://ondemand.hpc.wichita.edu/ and click “Files” at the top, you can easily upload and download files to and from Beoshock.

SFTP

You can configure an SFTP client using the same parameters you used to log in.

curl and wget

Both curl and wget are available without needing to load any modules. You can use them to download files from the internet, or from a web server you have access to.