Skip to content

batfish/af19-demo

master
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

This repository contains all of the code for the Batfish enabled CI pipeline demonstration from Ansiblefest 2019, including a script to initialize the Gitlab repository from where the CI tests are run.

Repository organization and CI workflow

Each time a commit is made to the Gitlab repository, device configs are generated and the generated configs are evaluated against defined policies.

  • The device configs are generated using jinja2 templates in the templates folder and input data in the inputs folder. The code for configuration generation is in the code folder.
  • Batfish supports both Ansible and Python (pytest) based policies. The policies folder contains pytest-based policies, and the ansible-policies folder contains Ansible-based policies. The setup for pytest policies (conftest.py) initializes a Batfish snapshot with generated device configs. Ansible policy playbooks assume that the snapshot has already been initialized.

The exact sequence of commands to generate configs and run policies are in template.gitlab-ci.yml file, which is uploaded to the Gitlab repository as its pipeline file (.gitlab-ci.yml) when the repository is initialized.

Network changes are proposed using Ansible playbooks in the playbooks folder. These playbooks change the inputs used to generate device configs. When these changes are committed to Gitlab, new configurations are generated and evaluated as described above.

Pre-requisites for running the demo

  • Access to Gitlab server and runner. See below for local setup.
    • In Gitlab, create a new project
      • Recommend naming it af19-template
    • Register the runner with the project
  • BF server running on localhost
  • Clone (or Download) this repository onto your local machine
  • Install python dependencies listed in requirements.txt file
    • pip install -r requirements.txt

Note: You can reduce execution time of the demo by leveraging an existing virtual environment that has the necessary dependencies install. To leverage an existing environment, edit the before_script stage of the Gitlab-CI pipeline file template.gitlab-ci.yml.

Local Gitlab setup on a Mac

The instructions below assume that you are installing a Gitlab server locally. If you use the online service, you can skip this part.

Gitlab server

Make sure that you have Docker installed on your laptop

  • Run the docker image
docker run --detach \
  --publish 443:443 --publish 80:80 --publish 22:22 \
  --hostname localhost \
  --name gitlab \
  --restart always \
  --volume ${HOME}/gitlab/config:/etc/gitlab \
  --volume ${HOME}/gitlab/logs:/var/log/gitlab \
  --volume ${HOME}/gitlab/data:/var/opt/gitlab \
  gitlab/gitlab-ce:latest

The initialization process may take a long time. You can track this process with the command sudo docker logs -f gitlab

  • Navigate to http://localhost/
    • Create an admin password

    • Register a new user

    • Create a new project

      • AF19-Template

      Initialize a README, but leave them private

    • Add an SSH key to your profile so you can clone the repos

      • Via user icon on top right -> settings -> ssh key
Clone AF19-Template repo locally

The demo requires a local clone of the GitLab repository that you just created.

Steps:

  • Go to the project page
  • Click on Clone (towards the top right)
    • Copy the Clone with SSH URL to clipboard

  • Open a terminal window
    • Type git clone git@localhost:samir-demo/af19-template.git (replace the git@... with what you copied from the Project page)
    • All of the playbooks referenced in the demo below, will need to be executed from the directory of this local clone.
Gitlab runner
Install GitLab runner
sudo curl --output /usr/local/bin/gitlab-runner https://gitlab-runner-downloads.s3.amazonaws.com/latest/binaries/gitlab-runner-darwin-amd64
sudo chmod +x /usr/local/bin/gitlab-runner
Register Gitlab runner with the project

This requires getting a token from Gitlab for the project. Steps:

  • Go to the project page

  • Settings (left side menu) —> CI/CD -> Expand Runners

  • Run gitlab-runner register

    • Enter the URL: http://localhost/
    • Enter the project specific Gitlab token from above
    • Accept default value for description
    • Accept the default value for tags
    • Enter shell as the executor
Start the Gitlab runner
  • gitlab-runner run if the process is not already running
  • gitlab-runner restart if the process is already running

Setup Gitlab Repo

Set the following environment variable:

  • GIT_TEMPLATE - This is the Git Clone URL for the repo you setup in Gitlab
    • Example: git@localhost:samir-demo/af19-template.git

Then, initialize your Gitlab repo by running the following command from the local clone of the batfish/af19-demo repository:

bash setup-gitlab-repo.sh

NOTE: This will reset the GitLab repository to the base state, so you can run the demo scenarios again.

Running the demos

To run the demos you will need to have a local clone of the af19-template repo. Both demo scenarios require you to run an Ansible playbook from this directory

Demo Scenario #1

The first demo involves adding a new leaf to the existing DC fabric. Run the following command:

ansible-playbook -i playbooks/inventory playbooks/add_leaf.yml

Enter the following information when prompted:

  • Enter the number for the new leaf router: 90
  • Enter the BGP AS number for the new leaf router: 65009
  • Enter commit message [adding leaf90 to fabric]:
    • this will accept the default commit message

And that's all you need to do to start the demo. In about 2-3 minutes you will see a new snapshot in Batfish.

Navigate to the Gitlab project AF19-Template to see the pipeline execution logs.

  • The build stage of the pipeline builds all of the configurations.
  • The test phase uploads the configurations to the BF server and evaluates the policies.
    • This is where you will see the results of the BF policy execution, under policy_eval

You should see 2 failures:

  • Unique BGP AS
  • All leaf routers have all host subnets

The leaf routers do not have bgp allow-as in configured, so leaf09 and leaf90 reject either others routes.

To correct the error, re-run the ansible playbook with the correct BGP ASN - 65090.

  • For correct operation each leaf should set the BGP AS to be equal to it's id/number

Demo Scenario #2

The first demo involves adding a new leaf to the existing DC fabric. Run the following command:

ansible-playbook -i playbooks/inventory playbooks/request_service.yml

Enter the following information when prompted:

  • Direction of request flow. IN for opening up access to internal service from the internet. OUT for opening up access to external service: IN
  • Enter the destination IP address/prefix: 10.100.10.0/24
  • Enter the source IP address/prefix: 0.0.0.0/0
  • Enter the IP protocol (TCP or UDP): tcp
  • Enter the destination port: 443
  • Enter the name of the application: SSL
  • Enter the change request id: CHG12345
  • Enter commit message [Change request CHG123345 for access to 10.100.10.0/24 application SSL from 0.0.0.0/0]:
    • this will accept the default commit message

And that's all you need to do to start the demo. In about 2-3 minutes you will see a new snapshot in the BFE Dashboard with policy execution complete.

Navigate to the Gitlab project AF19-Template to see the pipeline execution logs.

  • The build stage of the pipeline builds all of the configurations.
  • The test phase uploads the configurations to the BF server and evaluates the policies.
    • This is where you will see the results of the BF policy execution, under policy_eval

You should see 1 test failure:

  • Protect internal services (TCP)
    • Private subnets (new-TCP) reachability

This test fails because the network 10.100.0.0/16 has been designated as internal only. So traffic from the internet is allowed to it.

NOTE: If you want to re-run the scenario and select another subnet, you must provide a new source IP prefix and application since Capirca doesn't allow multiple entries for the same named object even if the definition is identical

Restart demo

To restart the demo, go back to the window where you ran the demo setup script and re-run the script.

About

Batfish Network CI demo for Ansiblefest 2019

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published