Scantron

Overview

Scantron is a distributed nmap scanner comprised of two components. The first is a Master node that consists of a web front end used for scheduling scans and storing nmap scan targets and results. The second component is an agent that pulls scan jobs from Master and conducts the actual nmap scanning. A majority of the application's logic is purposely placed on Master to make the agent(s) as "dumb" as possible. All nmap target files and nmap results reside on Master and are shared through a network file share (NFS) leveraging SSH tunnels. The agents call back to Master periodically using a REST API to check for scan tasks and provide scan status updates.

Scantron is coded for Python3.6+ exclusively and leverages Django for the web front-end, Django REST Framework as the API endpoint, PostgreSQL as the database, and comes complete with Ubuntu-focused Ansible playbooks for smooth deployments. Scantron has been tested on Ubuntu 18.04 and may be compatible with other operating systems. Scantron's inspiration comes from:

• dnmap
• Minions
• rainmap svn / rainmap github
• rainmap-lite

Scantron relies heavily on utilizing SSH port forwards (-R / -L) as an umbilical cord to the agents. Either an SSH connection from Master --> agent or agent --> Master is acceptable and may be required depending on different firewall rules, but tweaking the port forwards and autossh commands will be necessary. If you are unfamiliar with these concepts, there are some great overviews and tutorials out there:

• https://help.ubuntu.com/community/SSH/OpenSSH/PortForwarding
• https://www.systutorials.com/39648/port-forwarding-using-ssh-tunnel/
• https://www.everythingcli.org/ssh-tunnelling-for-fun-and-profit-autossh/

Use cases

Scantron is not engineered to be quickly deployed to a server to scan for a few minutes, then torn down and destroyed.
It's better suited for having a set of static scanners (e.g., "internal-scanner", "external-scanner") with a relatively static set of assets to scan.

Architecture Diagram

Hardware Requirements

• Agent: If you plan on compiling masscan on an agent, you'll need at least 1024 MB of memory. It fails to build with only 512 MB. If you do not want to build masscan, set install_masscan_on_agent to False in ansible-playbooks/group_vars/all

• Master: 512 MB of memory was the smallest amount successfully tested.

Ansible Deployment Server and Initial Setup

This is your local box, preferably Linux. Ansible >= 2.4.0.0 is the minimum version required for utilizing ufw comments.

Clone the project and execute initial_setup.sh.

# Clone scantron project.
git clone https://github.com/rackerlabs/scantron.git
cd scantron
./initial_setup.sh  # Run as non-root user.

Installation

Installation requires a general knowledge of Python, pip, and Ansible. Every attempt to make the deployment as simple as possible has been made.

Cloud Provider Caveats

NAT'd instances

If the Master server is actually a RFC1918 IP and not the public IP (because of NAT), the NAT'd RFC1918 IP (e.g., 10.1.1.2) will have to be added to the ALLOWED_HOSTS in ansible-playbooks/roles/master/templates/production.py.j2

This is common in AWS and GCP environments.

IBM Cloud

Per https://github.com/0xtavian: For the Ansible workload to work on IBM Cloud, edit the file /boot/grub/menu.lst by changing

# groot=LABEL...

to

# groot=(hd0)

Update hosts

Edit the hosts in this file:

• ansible-playbooks/hosts

Master Installation

The recommendation is to deploy Master first.

Update Master Ansible Variables

Edit any variables in these files before running playbook:

• ansible-playbooks/group_vars/all

If you plan on utilizing the same API key across all agents (not recommended, but easier for automated deployments), change utilize_static_api_token_across_agents to True. This prevents you from having to log into each agent and update agent_config.json with the corresponding API key. The group_vars/static_api_key will be created by the Master ansible playbook. The Ansible agent playbook will autofill the agent_config.json.j2 template with the API key found in group_vars/static_api_key.

WARNING: The agent_config.json.j2 will generate a random scan_agent (e.g., agent-847623), so if you deploy more than 1 agent, you won't run into complications with agent name collisions. You will, however, need to add create the user on Master, since Master returns scheduled jobs to the agent based off the agent's name!

Rename master/scantron_secrets.json.empty to master/scantron_secrets.json (should be done for you by initial_setup.sh)

Update Master Secrets Variables

Update all the values master/scantron_secrets.json if you do not like ones generated using initial_setup.sh. Only the production values are used.

• All Scantron Django passwords have a minimum password length of 12.

• For the "SECRET_KEY", per Django's documentation: The secret key must be a large random value and it must be kept secret.

Change scantron user password (optional)

The scantron operating system user password is not really leveraged and is populated by providing a salted hash of a random password generated using Python's passlib library. If you want to change the password, you will have to generate a hash for the desired password and update the temp_user_pass variable in scantron/ansible-playbooks/roles/add_users/vars/main.yml.

pip3 install passlib

python3 -c "from passlib.hash import sha512_crypt; import getpass; print(sha512_crypt.encrypt(getpass.getpass()))"

Execute Master Ansible Playbook

Ensure you have a SSH key (or username/password) to access the Master box, specified by --private-key in the Ansible command. User must also have password-less sudo privileges.

cd ansible-playbooks

# non-root user with password-less sudo capabilities.
ansible-playbook master.yml -u ubuntu --become --private-key=<agent SSH key>

# root user.
ansible-playbook master.yml -u root --private-key=<agent SSH key>

Change Django user passwords with manage.py (optional)

cd into the master directory scantron/master and run the following to change the admin (or whatever user needs their password changed) user password.

python3 manage.py changepassword admin

Agent Installtion

Update Agent Ansible Variables

Edit any variables in these files before running playbook:

• ansible-playbooks/group_vars/all
• ansible-playbooks/roles/agent/vars/main.yml

Ensure proper user permissions

Ensure you have a SSH key (or username/password) to access the agent box, specified by --private-key in the Ansible command. The user must also have password-less sudo privileges. If you are creating the boxes on AWS, then the user is ubuntu for Ubuntu distros and the user already has password-less sudo capabilities. If you need to add password-less sudo capability to a user, create a /etc/sudoder.d/<USERNAME> file, where <USERNAME> is the actual user, and populate it with:

<USERNAME> ALL=(ALL) NOPASSWD: ALL

SSH-ing in as root will also work for the Ansible deployment, but is not generally recommended.

Execute Agent Ansible Playbook

cd ansible-playbooks

# non-root user with password-less sudo capabilities.
ansible-playbook agent.yml -u ubuntu --become --private-key=<agent SSH key>

# root user.
ansible-playbook agent.yml -u root --private-key=<agent SSH key>

Adding additional agents

A Scantron agent is synonymous with a user.

Agents <--> Users

Users / agents are added through the webapp, so once a user / agent is added, an API token is automatically generated for that user / agent. The user's / agent's password is not necessary for Scantron to function since all user / agent authentication is done using the API token. The username and password can be used to login to the webapp to test API functionality. More API testing information can be found in the Test Agent API section of this README.

Update /etc/rc.local with agent IPs for autossh

This is done automatically for one agent through Ansible. You may have to add additional lines and update SSH keys for each agent if they are different. These commands are for Master connecting to the agents.

In this example:

• Master - 192.168.1.99
• agent1 - 192.168.1.100
• agent2 - 192.168.1.101

# Master --> Agent 1
su - autossh -s /bin/bash -c 'autossh -M 0 -f -N -o "StrictHostKeyChecking no" -o "ServerAliveInterval 60" -o "ServerAliveCountMax 3" -p 22 -R 4430:127.0.0.1:443 -R 2049:127.0.0.1:2049 -i /home/scantron/master/autossh.key [email protected]'

# Master --> Agent 2
su - autossh -s /bin/bash -c 'autossh -M 0 -f -N -o "StrictHostKeyChecking no" -o "ServerAliveInterval 60" -o "ServerAliveCountMax 3" -p 22 -R 4430:127.0.0.1:443 -R 2049:127.0.0.1:2049 -i /home/scantron/master/autossh.key [email protected]'

If Master cannot SSH to an agent, then the autossh command will be run on the agent and the port forwards will be local (-L) instead of remote (-R).

# Master <-- Agent 1
su - autossh -s /bin/bash -c 'autossh -M 0 -f -N -o "StrictHostKeyChecking no" -o "ServerAliveInterval 60" -o "ServerAliveCountMax 3" -p 22 -L 4430:127.0.0.1:443 -L 2049:127.0.0.1:2049 -i /home/scantron/master/autossh.key [email protected]'

Agents

Agent's agent_config.json

agent_config.json is a configuration file used by agents to provide basic settings and bootstrap communication with Master. Each agent can have a different configuration file.

The "api_token" will have to be modified on all the agents after deploying Master!

Agent settings:

scan_agent: Name of the agent. This name is also used in the agent's HTTP User-Agent string to help identify agents calling back in the nginx web logs.

api_token: Used to authenticate agents. Recommend different API Tokens per agent, but the same one could be used.

master_address: Web address of Master. Could be 127.0.0.1 if agent traffic is tunneled to Master through an SSH port forward.

master_port: Web port Master is listening on.

callback_interval_in_seconds: Number of seconds agents wait before calling back for scan jobs.

number_of_threads: Experimental! Number of threads used to execute scan jobs if multiple jobs may be required at the same time. Keep at 1 to avoid a doubling scanning race condition.

target_files_dir: Name of actual agent target_files directory on the agent box.

scan_results_dir: Name of actual agent scan_results directory on the agent box.

log_verbosity: Desired log level for logs/agent.log

# Level     Numeric value
# CRITICAL  50
# ERROR     40
# WARNING   30
# INFO      20
# DEBUG     10

http_useragent: HTTP User-Agent used instead of nmap's default Mozilla/5.0 (compatible; Nmap Scripting Engine; https://nmap.org/book/nse.html).

Agent Execution

Update all the agents' agent_config.json files with their respective api_token for the agent by logging in as admin and browsing to https://<HOST>/scantron-admin/authtoken/token to see the corresponding API token for each user / agent.

Option 1: Run agent as a service

Enable scantron-agent service at startup.

systemctl daemon-reload  # Required if scantron-agent.service changed.
systemctl enable scantron-agent

Disable scantron-agent service at startup.

systemctl disable scantron-agent

Scantron service troubleshooting commands.

systemctl status scantron-agent
systemctl start scantron-agent
systemctl stop scantron-agent
systemctl restart scantron-agent

Option 2: Run agent as standalone script

Use screen to avoid the script dying after disconnecting through SSH.

screen -S agent1  # Create a screen session and name it agent1, if using screen.

cd agent
source .venv/bin/activate
python agent.py -c agent_config.json

CTRL + a + d  # Break out of screen session, if using screen.
screen -ls  # View screen job, if using screen.

screen -r agent1  # Resume named screen session, if using screen.

Agent Troubleshooting

Verify SSH connection from Master with reverse port redirect is up on each agent. Any traffic hitting 127.0.0.1:4430 will be tunneled back to Master. This port is for communicating with the API. Any traffic hitting 127.0.0.1:2049 will connect back to the NFS share on Master.

tcp    0    0 127.0.0.1:4430    0.0.0.0:*    LISTEN    1399/sshd: autossh
tcp    0    0 127.0.0.1:2049    0.0.0.0:*    LISTEN    1399/sshd: autossh

Check each agent's root crontab to ensure nfs_watcher.sh is being run every minute.

crontab -l -u root

Test Agent API

If you need to test the API without running the agent, ensure there is a 'pending' scan set to start earlier than the current date and time. The server only returns scan jobs that have a 'pending' status and start datetime earlier than the current datetime.

# Not using SSH tunnels.
curl -k -X GET -H 'Authorization: Token <VALID API TOKEN>' https://192.168.1.99:443/api/scheduled_scans

# Using SSH tunnels.
curl -k -X GET -H 'Authorization: Token <VALID API TOKEN>' https://127.0.0.1:4430/api/scheduled_scans

You can also log into the webapp using the agent name and password and browse to /api/?format=json to view any scan jobs. The username and agent name are the same from the webapp's point of view.

Master

Master target_files Folder

• Place files with target IPs/hosts (fed to nmap -iL switch) in master/target_files/
• target_files is an NFS share on Master that the agent reads from through an SSH tunnel.

Master scan_results folder

• nmap scan results from agents go here.
• master/scan_results/ is an NFS share on Master that the agent writes to through an SSH tunnel.

Master Troubleshooting

1). Ensure SSH tunnels setup in /etc/rc.local are up.

netstat -nat | egrep "192.168.1.100|192.168.1.101"
ps -ef | egrep autossh

2). Django logs can be found here: /var/log/webapp/django_scantron.log

3). Check nginx logs for agent name in User-Agent field to determine which agents are calling home. nginx logs: tail -f /var/log/nginx/{access,error}.log

4). uwsgi logs: /home/scantron/master/logs

Known issues with Master NFS share

If you need to reboot a box, do it with the provided clean_reboot.sh script that will stop all relevant services. Without stopping the nfs-kernel-server service gracefully, sometimes the OS will hang and get angry.

Miscellaneous

Updating nmap version

Ubuntu's nmap version pulled using apt is fairly out-of-date and the recommendation for Scantron's agents is to pull the latest version.

For RPM-based Distributions, the latest .rpm packages can be found here https://nmap.org/dist/?C=M&O=D. However, for Debian-based distributions, you must utilize alien to convert the .rpm to a .deb file https://nmap.org/book/inst-linux.html or compile from source. Recommend going down the alien route before compiling from source.

alien

VERSION=7.70-1  # CHANGE THIS TO LATEST VERSION

apt install alien -y
wget https://nmap.org/dist/nmap-$VERSION.x86_64.rpm
alien nmap-$VERSION.x86_64.rpm

apt remove nmap -y
apt remove ndiff -y
dpkg --install nmap_*.deb

Compile nmap from source

Another option is to compile nmap from source. This is dynamically compiled and must be done on the box where nmap is going to be run from. Note that past experience had a compiled nmap version returning a different banner than the provided apt version...so your mileage may vary.

VERSION=7.70-1  # CHANGE THIS TO LATEST VERSION

wget https://nmap.org/dist/nmap-$VERSION.tar.bz2
bzip2 -cd nmap-$VERSION.tar.bz2 | tar xvf -
cd nmap-$VERSION
./configure --without-ncat --without-ndiff --without-nmap-update --without-nping --without-subversion \
--without-zenmap --with-libdnet=included --with-libpcap=included --with-libpcre=included
make
./nmap -V

Get nmap --top-ports

This provides a list of the actual ports being scanned when the --top-ports option is used:

# TCP
nmap -sT --top-ports 1000 -v -oG -

# UDP
nmap -sU --top-ports 1000 -v -oG -

Sorted list based on frequency.

# TCP sorted list based on frequency.
egrep /tcp /usr/share/nmap/nmap-services | sort -r -k3

# UDP sorted list based on frequency.
egrep /udp /usr/share/nmap/nmap-services | sort -r -k3

Source: https://security.stackexchange.com/questions/78618/is-there-a-nmap-command-to-get-the-top-most-common-ports

Workflow

1. Create user/agent. By default, Ansible creates agent1.

   

2. Create nmap command

   

3. Create a site

   • IPs, IP subnets, and FQDNs are allowed.
   • IP ranges (192.168.1.0-10) are not currently supported.
   • The targets and excluded_targets are validated using master/extract_targets.py, which can also be used as a stand alone script.

   

4. Create scan

   • Select start time
   • Add start date
   • Add recurrence rules (if applicable)

   The /home/scantron/master/scan_scheduler.sh cronjob checks every minute to determine if any scans need to be queued. If scans are found, it schedules them to be picked up by the agents.

   

5. View currently executing scan results

   cd /home/scantron/master/scan_results/pending
   ls -lart

   Completed scans are moved to the /home/scantron/master/scan_results/completed directory.

6. Process scans

   Scan files are moved between a few folders.

   /home/scantron/master/scan_results/pending - Pending scan files from agents are stored here before being moved to scan_results/complete

   /home/scantron/master/scan_results/complete - Completed scan files from agents are stored here before being processed by nmap_to_csv.py

   The scantron user executes a cron job (nmap_to_csv.sh which calls nmap_to_csv.py) every 5 minutes that will process the .xml scan results found in the complete directory and move them to the processed directory.

   /home/scantron/master/scan_results/processed - nmap scan files already processed by nmap_to_csv.py reside here.

   /home/scantron/master/for_bigdata_analytics - csv files for big data analytics ingestion

API Documentation

There are 3 ways to explore and play around with the API. The first is the Django REST Framework view:

You can also dig through the API documentation using ReDoc:

Lastly, you can interact with the API using Swagger:

Database Model Graph

Generated using django-extensions's graph_models.

Contributing

If you would like to contribute, please adhere to the Python code black formatter rules specifying a line length of 120.

More information about black can be found here (https://github.com/ambv/black)

Robot Image

Robot lovingly delivered by Robohash.org (https://robohash.org)