Skip to content

Latest commit

 

History

History
149 lines (97 loc) · 7.98 KB

README.md

File metadata and controls

149 lines (97 loc) · 7.98 KB
Raw

Platform Hub – API Server & Backend

Tech summary

A Ruby on Rails 5.0 API-only stack to provide a mainly JSON based API and backend for business logic and persistence, and for communicating with other backend and external services. Using PostgreSQL for persistence of data.

Dev

Prerequisites

Ruby v2.3.8

If you're using RVM or some other Ruby manager, it may pick up the .ruby-version file automatically and use / set up the correct version of Ruby for you. If not, you'll need to install Ruby v2.3.8.

Package Dependencies

The following packages are required to run the API (may vary across Operating Systems):

  • Ruby v2.3.8. If you're using RVM or some other Ruby manager, it may pick up the .ruby-version file automatically and use / set up the correct version of Ruby for you. If not, you'll need to install Ruby v2.3.8 or above (included in the below list).
  • Developer Packages:
    • Debian/Ubuntu: sudo apt-get install ruby ruby-dev postgresql-client libpq-dev
    • Fedora/CentOS: sudo yum install rubygems ruby-devel postgresql-client postgresql-devel
    • OSX: brew install ruby ruby-build postgresql

Bundler

Bundler is used for dependency management. Install this with gem install bundler.

PostgresSQL v12.5

For local development, we recommend using the scripts provided in ./bin/postgres/* to easily manage a locally running PostgreSQL container with the correct config set up already. See the relevant README doc for more details.

ElasticSearch v5.5.1

For local development, we recommend using the scripts provided in ./bin/elasticsearch/* to easily manage a locally running ElasticSearch container with the correct config set up already. See the relevant README doc for more details.

Dev flows

Set up local config

All config is passed as environment variables. The .env file provided is loaded automatically by the Rails environment and contains some config options for an "out-of-the-box" developer experience (note: this file is committed to the Git repo).

You can override any config value by adding it to a .env.local file that you will need to create (this won't and shouldn't get committed to the Git repo). You must have certain secret values set up in this .env.local file to make things work locally. You may need to ask a fellow developer for some of these values. The following are the variables that need to be set in this file:

  • SECRET_KEY_BASE – a string used for encryption. Usually 128 bytes.
  • GITHUB_CLIENT_ID – the client ID for the GitHub OAuth app used for the identity connection. Make sure this is a test one and not a production one!
  • GITHUB_CLIENT_SECRET – the client secret for the GitHub OAuth app used for the identity connection. Make sure this is a test one and not a production one!
  • AGENT_GITHUB_TOKEN – the access token used for GitHub onboarding flows. This token MUST be for a user that has owner access to the GitHub organisation specified below.
  • AGENT_GITHUB_ORG – the GitHub organisation that users will be onboarded on to.
  • AGENT_GITHUB_ORG_MAIN_TEAM_ID – the integer ID of the GitHub team that users will be onboarded on to.
  • AGENT_GITLAB_BASE_URL – the base URL of the GitLab instance managed by the hub.
  • AGENT_GITLAB_TOKEN – the personal access token to access the API of the abovementioned GitLab instance.
  • AGENT_KEYCLOAK_CLIENT_ID - Keycloak client ID. Make sure this is a test one and not a production one!
  • AGENT_KEYCLOAK_CLIENT_SECRET - Keycloak client secret. Make sure this is a test one and not a production one!
  • AGENT_KEYCLOAK_USERNAME - Keycloak agent username. This is a Keycloak user with sufficient privileges to manage users in AGENT_KEYCLOAK_REALM.
  • AGENT_KEYCLOAK_PASSWORD - Keycloak agent password.
  • AGENT_KEYCLOAK_BASE_URL - Keycloak base URL. Make sure this is not a production Keycloak URL!
  • AGENT_KEYCLOAK_REALM - The Keycloak realm in which users will be managed.
  • SLACK_WEBHOOK – the webhook URL for Slack integration – use wisely when developing locally, just in case unwanted messages get sent to a public Slack channel – set to noop or some other string to ensure nothing gets sent out.
  • NOTIFY_EXCEPTIONS – specifies whether unhandled exceptions should be notified on Slack. Either omit this entirely, or set to false, to disable.
  • NOTIFY_EXCEPTIONS_SLACK_CHANNEL – the specific Slack channel to send notifications about unhandled exceptions.

Note that these env files only work for local development and testing.

Install dependencies

bundle

Set up your local db

bin/rails db:setup

This will also set up your local tests database used when running the test suite.

Common tasks

  • bin/rails server – runs a local server to serve the API (Ctrl+C to stop)
  • bin/rails jobs:work – runs a non-daemonised delayed_job worker to process background jobs (Ctrl+C to stop)
  • bundle exec rspec – runs the tests (specified in the /spec folder)
  • bin/rails console – runs a local Rails console to access your app
  • bin/rails db:migrate – runs database migrations to get your local database to the latest schema

See the Rails CLI guide for details on many more useful command line tasks, for managing migrations, generating files, etc.

General architectural points of interest

  • As much as possible, we try to break out coherent, well isolated and single responsibility tasks/logic into "services" in the app/services folder.
  • The services in app/services/agents/* are intended to provide "root" access to external services like GitHub, and thus should be used cautiously and be well tested wherever used!

Announcements delivery

Announcements that need delivering (email, Slack, etc.) are processed in a background job (using delayed_job). The processor job itself needs to be manually triggered using the following (this allows us to run it on a schedule):

bin/rails announcements:trigger_processor_job

… this puts a job on the queue. To work through the jobs in the queue, start up a worker:

bin/rails jobs:work

Slack delivery

As long as you have a valid SLACK_WEBHOOK set in your .env.local (or passed in through your environment), delivery of announcements to Slack channels will work. Be careful when using this in local development though – only send to Slack channels that are okay to receive test messsages.

Email delivery

To set up local email delivery, ensure you have the following env variables set (either in .env.local or passed through your environment):

  • EMAIL_SMTP_ADDRESS
  • EMAIL_SMTP_PORT
  • EMAIL_SMTP_USERNAME
  • EMAIL_SMTP_PASSWORD

As usual, be careful when sending out emails from your local development environment!

Help search

The help search feature uses ElasticSearch; most of the relevant functionality is in the HelpSearchService class. The global instance HelpSearchService.instance can be used.

To reindex all help items from scratch to a fresh index:

HelpSearchService.instance.reindex_all force: true

To reindex all help items into an existing index:

HelpSearchService.instance.reindex_all

Troubleshooting

Test database issues

If you experience issues with your tests starting up due to a database error (e.g. a ActiveRecord::NoEnvironmentInSchemaError that often occurs after migrating to a newer schema), then recreate the whole test db with:

bin/postgres/console -c "drop database phub_test"
bundle exec rake db:setup

Note: make sure your development db is fully up to date before running this (i.e. make sure all migrations have run, with bin/rails db:migrate) so that the test db has the latest schema.