Skip to content

Latest commit

 

History

History
196 lines (141 loc) · 8.04 KB

installation.md

File metadata and controls

196 lines (141 loc) · 8.04 KB
Raw

HubCommander Installation

The steps below make the following assumptions:

  • You are using Linux or macOS
  • Have Python 3.5 installed
  • Have a GitHub organization to manage
  • Have a GitHub user with owner privileges.
  • Have Slack API credentials
  • Have a Slack Channel dedicated for running the bot in
  • Preferably have a means of protecting secrets -- you will need to write some code here!

Basic Installation

HubCommander is dependent on the slackhq/python-rtmbot release 0.4.0). Please review details about the python-rtmbot before continuing to install HubCommander.

The python-rtmbot typically operates by placing a plugin in the plugins directory.

A docker image for HubCommander is provided to help get up and running as fast as possible. Alternatively, a shell script is provided here that will fetch the rtmbot, and will git clone HubCommander into the plugins directory.

Once that is done, you will need to perform all additional configuration steps required to make it function in your environment, including credential management.

Install the bot

For Docker:

Continue reading this document first. Once done, continue reading the Docker details (linked below).

For macOS:

  1. Install Homebrew
  2. Install Python 3.5+: brew install python3
  3. Proceed to "Continued Instructions"

For Ubuntu or other Linuxes:

  1. Run sudo apt-get update
  2. Run sudo apt-get install python3 python3-venv curl git -y
  3. Proceed to "Continued Instructions"

Continued Instructions

  1. Clone the HubCommander git repository: git clone [email protected]:Netflix/hubcommander.git.

  2. Run the following commands:

    chmod +x hubcommander/basic_install.sh
./hubcommander/basic_install.sh

  3. Proceed to "Configuration"

OPTIONAL for unit tests:

If you are installing for development you will need to install the testing dependencies. Follow the Install the unit test requirements section of the contributing guide for details.

Configuration

The primary steps for configuration center around credential management for Slack, and GitHub (and optionally Travis CI and Duo if you utilize those services).

Decrypting secrets

Out of the box, HubCommander is configured to receive secrets from environment variables. This is provided to simplify running HubCommander in Docker. These secrets should not be stored at rest unencrypted.

HubCommander also provides an AWS KMS method for extracting an encrypted JSON blob that contains the secrets.

If your organization utilizes a different mechanism for encrypting credentials, you will need to add code to decrypted_creds.py's get_credentials() function. Please DO NOT store credentials and tokens in plaintext here!

No matter the encryption mechanism utilized, all secrets are passed into each plugin's setup() method, which enable the plugins to make authenticated calls to their respective services. You must get the credentials required for plugins to work.

  1. Contact your Slack administrator to obtain a Slack token that can be used for bots.
  2. If you haven't already, create a GitHub bot account, and invite it to all the organizations that you manage with the owner role. Also, please configure this account with a strong password and 2FA! (Don't forget to back up those recovery codes into a safe place!)

GitHub Configuration

Each plugin in HubCommander typically has a config.py file. This is where you can place in any additional configuration that the plugin supports.

For the GitHub plugin, you are required to define a Python dict with the organizations that you manage. An example of what this dict looks like can be found in the sample github/config.py file.

At a minimum, you need to specify the real name of the organization, a list of aliases for the orgs (or an empty list), whether the organization can only create public repos (via the public_only boolean), as well as a list of dicts that define the teams specific to the organization for new repositories will be assigned with. This dict consists of 3 parts: the id of the GitHub org's team (you can get this from the list_teams GitHub API command, along with the permission for that team to have on newly created repos (either pull, push, or admin), as well as the actual name of the team.

GitHub Configuration: API Token

Once you have a GitHub bot user account available, it is time to generate an API token. This will be used by HubCommander to perform privileged GitHub tasks.

You will need to create an access token. To do this, you will need to:

  1. Log into your GitHub bot user's account.

  2. Visit this settings page to see the Personal Access Tokens

  3. Click Generate new token.

  4. Provide a description for this token, such as HubCommander Slack GitHub Bot API token.

  5. Provide the following scopes:

    • repo (All)
    • read:org
    • write:org
    • delete_repo
    • (These scopes can be later modified)

  6. Click Generate Token to get the API key. This will only be displayed once. You can always re-generate a new token, but you will need to modify your HubCommander configuration each time you do.

HubCommander Secrets

Once you have your GitHub and Slack Tokens, you are now ready to configure HubCommander (if you wish to make use of Travis CI and Duo integration, please refer to the docs for those plugins here and here).

You will need to encrypt the Slack and GitHub credentials. If you make use of AWS, a KMS example is provided in decrypted_creds.py.

At a minimum, HubCommander requires the following secrets (as a Python dict):

{
    "SLACK": "Your slack token here...",
    "GITHUB": "Your GitHub API token here..."
}

Encrypt this via any desirable means you choose, and add in your decryption code to the get_credentials() function.

Note: You will need to ensure that the rtmbot's Slack credentials are also encrypted (this requires the same Slack token). Use whatever deployment mechanism you have in place to ensure that it is encrypted and in a safe place before running the application.

Additional Command Configuration

Please refer to the documentation here for additional details.

Running HubCommander

rtmbot Configuration

Regardless of how you run the bot, you will need to worry about the rtmbot configuration. The factory docker image generates this dynamically, but if you decide to make changes to the image, you will need to be aware of how this works.

The rtmbot.conf file is required to be placed in the base directory of the rtmbot. This file MUST have some elements in it. Namely, it must look similar to this:

DEBUG: True
SLACK_TOKEN: "YOUR-SLACK-TOKEN-HERE"
ACTIVE_PLUGINS:
    - hubcommander.hubcommander.HubCommander

Using Docker

Continue reading the HubCommander Docker documentation here.

Not using Docker

If you ran the installation shell script, and made all the configuration file changes you need, then you are ready to run this!

You will simply follow the instructions for running the python-rtmbot, which is typically to run:

# Activate your venv:
source /path/to/venv/.../bin/activate
rtmbot

If all is successful, you should see no errors in the output, and your bot should appear in the Slack channel that it was configured to run in.

Test that it works by running !Help, and !ListOrgs.