Skip to content

Latest commit

 

History

History
81 lines (48 loc) · 4.2 KB

configuration.md

File metadata and controls

81 lines (48 loc) · 4.2 KB

Configuration

This document explains the various configurable parts of the Ruby Toolbox app that need to be set up for a production deployment

PostgreSQL (required)

A PostgreSQL database is required to store data. It's recommended to provide it via the standard DATABASE_URL environment variable.

Redis (required)

A Redis instance is needed for sidekiq background processing. It is recommended to configure it by providing a REDIS_URL environment variable.

Github API Client (required)

The GITHUB_TOKEN is used by the GraphQL API client to fetch repository data. Since Github's GraphQL API does not support unauthenticated usage you must provide this, even in local development.

A "Personal Access Token" will do just fine, you can create one by following Github's instructions. At the time of writing, the direct link to the corresponding settings page is https://github.com/settings/tokens

To keep things nicely separated the main production site has a dedicated rubytoolbox-bot user account for these tokens.

Maximum Threads (optional)

The RAILS_MAX_THREADS option sets the number of threads to be used per puma web server and sidekiq background worker process. It should be an integer number. A fallback default is pulled from the .env file.

Canonical Domain and SSL enforcement (optional)

Setting CANONICAL_HOST=www.ruby-toolbox.com will enforce any requests made to the app under a different domain to be redirected to the canonical one.

It will also enable SSL enforcement - if a user visits via plain HTTP, they will be redirected to the SSL variant.

Appsignal Push API Key (optional)

In order to use AppSignal for exception tracking and performance monitoring, the APPSIGNAL_PUSH_API_KEY must be set. It can be created from the appsignal dashboard.

Sidekiq Admin Dashboard (optional)

The sidekiq web UI is mounted at /ops/sidekiq. It is protected by the SIDEKIQ_PASSWORD using basic auth. The username is irrelevant.

If this is not configured the app generates a random password on boot.

Catalog Update Github Webhook (optional)

On every update the catalog builds a new JSON export and deploys it to Github Pages. In order to synchronize the live catalog as fast as possible, a Github event webhook can be configured to trigger catalog synchronization on successful catalog builds.

If this is not configured an hourly update cron job will pull the latest data, so this is not a required configuration option.

  • Payload URL: https://www.ruby-toolbox.com/webhooks/github
  • Content-Type: application/json
  • Secret: Some Secret String. Also set this as GITHUB_WEBHOOK_SECRET on the app environment.
  • ✔️ Enable SSL Verification
  • "Let me select individual events"
    • ✔️ Statuses
  • ✔️ Active

See also rubytoolbox#339

Serve assets from Rails (optional)

By setting RAILS_SERVE_STATIC_FILES to true the Rails app will be hosting the assets. The regular production app is running on Heroku and has this enabled by default, including asset precompilation that Heroku takes care of automatically.

Development Configuration

Autocorrect Rubocop offenses when running with Guard

Set AUTOCORRECT=true when running guard for continuouos testing to automatically fix any auto-correctable rubocop offenses that are found.

Run feature specs in real Chrome

By default capybara acceptance tests are run via Chrome Headless. By passing CHROME_DEBUG=true when running the tests the regular selenium_chrome driver will be used instead, allowing visual debugging.