This is a fork for the irassh SSH and Telnet Honeypot effort that allows to run it with docker compose. The docker-compose service will listen on the public system interface on port 22 for live usage.

Software required:

• Python 2.7+, (Python 3 not yet supported due to Twisted dependencies)
• python-virtualenv

For Python dependencies, see requirements.txt

• src/irassh.cfg - Cowrie's configuration file. Default values can be found in cowrie.cfg.dist
• src/data/fs.pickle - fake filesystem
• src/data/userdb.txt - credentials allowed or disallowed to access the honeypot
• src/dl/ - files transferred from the attacker to the honeypot are stored here
• src/honeyfs/ - file contents for the fake filesystem - feel free to copy a real system here or use bin/fsctl
• log/irassh.json - transaction output in JSON format
• log/irassh.log - log/debug output
• log/tty/*.log - session logs
• src/txtcmds/ - file contents for the fake commands
• bin/createfs - used to create the fake filesystem
• bin/playlog - utility to replay session logs

• Create 2 folders: log and log/tty
• bin/irassh start - start the server
• bin/irassh stop - stop the server
• Start client: ssh root@localhost, input any pwd
• Run playlog: bin/playlog log/tty/[file_name]

• Step 1: Run bin/irassh start with manual set to True in irassh/shell/honeypot.py and start bin/manual.py to give manual commands
• Step 2: After enough commands have been recorded a file named manual/input/cmd.p should appear. Use this file as an input for irassh/rl/manual.py ( e.g. python manual.py manual/input/cmd.p expertFE.p -p )
• Step 3: Use the policy obtained in the previous step in irassh/actions/proxy.py by setting the value of expertFE to it
• Step 4: Run bin/irassh start with manual set to False and useIRL set to True and input enough commands until the irl agent is trained. The result should be saved in a pickle file named using the behavior attribute of the irl_agent with "-optimal_weight.p" appended to it
• Step 5: Run irassh/rl/policy2reward.py with the output from the previous step to create a pickle file that contains the reward function for the q_learner (i.e. python policy2reward.py DefaultBehavior-optimal_weight.p cmd2number_reward.p
• Step 6: Train a Reinforcement Learning agent by starting irassh with manual and useIRL set to False. Make sure the cmd2number_reward is set correctly in irassh/actions/proxy.py

Just run sudo docker-compose build && sudo docker-compose up -d

• Turn on manual = true in irassh/shell/honeypot.sh
• Turn on server: bin/irassh start
• Turn on manual console: python bin/manual
• Connect to server: ssh root@localhost -p 22

• Add action to playlog
• Add action mysql log
• Move all functions from rassh to irassh