openlane

This repository houses the core server and orchestration elements which are at the heart of the openlane cloud service.

Features

At it's core, this repo is a collection of services built on top of an entity framework which allows us to:

• Model database schemas as graph structures
• Define schemas as programmatic go code
• Execute complex database queries and graph traversals easily
• Extend and customize using templates and code generation utilities
• Type-safe resolvers and GraphQL schema stitching
• Code generated audit / history tables for defined schemas

On top of this powerful core we also have an incredible amount of pluggable, extensible services:

• Authentication: we today support password, OAuth2 / Social login providers (Github, Google), Passkeys as well as standard OIDC Discovery flows
• Multi-factor: built-in 2FA mechanisms, TOTP
• Authorization: extensible and flexible permissions constructs via openFGA based on Google Zanzibar
• Session Management: built-in session management with JWKS key validation, encrypted cookies and sessions
• Robust Middleware: cache control, CORS, Rate Limiting, transaction rollbacks, and more
• Queuing and Scheduling: Task management and scheduling with riverqueue
• External Storage Providers: store data in AWS S3, Google GCS, or locally
• External Database Providers: Leverage NeonDB, or other PostgreSQL compatible vendors and libraries
• Data Isolation and Management: Hierarchal organizations and granular permissions controls

Development

Dependencies

The vast majority of behaviors of the system can be turned on or off by updating the configuration parameters found in config; in some instances, we've made features or integrations with third party systems which are "always on", but we're happy to receive PR's wrapping those dependencies if you are interested in running the software without them!

Installing Dependencies

Setup Taskfile by following the instructions and using one of the various convenient package managers or installation scripts. After installation, you can then simply run task install to load the associated dependencies. Nearly everything in this repository assumes you already have a local golang environment setup so this is not included. Please see the associated documentation.

Updating Configuration Settings

See the README in the config directory.

Starting the Server

1. Copy the config, this is in .gitignore so you do not have to worry about accidentally committing secrets

   cp ./config/config-dev.example.yaml ./config/.config.yaml

2. Update the configuration with whatever respective settings you desire; the defaults inside should allow you to run the server without a problem

3. Create the openlane.fga

   cp fga/model/model.fga fga/model/openlane.fga

4. Use the task commands to start the server

   Run the core server in development mode with dependencies in docker

   task run-dev

   Run fully in docker

   task docker:all:up

5. In a separate terminal, with the server running, you can create a verified test user by running:

   task cli:user:all

6. Once this command has finished ^, you can login and perform actions as user [email protected] with password mattisthebest1234

Creating Queries in GraphQL

The best method of forming / testing queries against the server is to run task docker:rover which will launch an interactive query UI.

If you are running the queries against your local repo, you will have CORS issues using the local running apollo. Instead, its recommended to use the apollo sandbox and ensure the following origin is allowed in your config/.config.yaml

server:
  cors:
    allowOrigins:
      - https://studio.apollographql.com

OpenFGA Playground

You can load up a local openFGA environment with the compose setup in this repository; task fga:up - this will launch an interactive playground where you can model permissions model(s) or changes to the models

Creating a new Schema

To ease the effort required to add additional schemas into the system a template + task function has been created. This isn't doing anything terribly complex, but it's attempting to ensure you have the minimum set of required things needed to create a schema - most notably: you need to ensure the IDMixin is present (otherwise you will get ID type conflicts) and a standard set of schema annotations.

NOTE: you still have to make intelligent decisions around things like the presence / integration of hooks, interceptors, policies, etc. This is saving you about 10 seconds of copy-paste, so don't over estimate the automation, here.

To generate a new schema, you can run task newschema -- [yourschemaname] where you replace the name within []. Please be sure to note that this isn't a command line flag so there's a space between -- and the name.

Migrations

We use atlas and goose to create and manage our DB migrations - you can trigger one via task atlas:create and that will generate the necessary migrations. There should be a new migration file created in db/migrations and db/migrations-goose-postgres. On every PR, the Atlas integration also creates comments with any issues related to the schema changes / migrations.

Deploying

The only "supported" method of deploying today is locally, but we have a WIP Helm chart which can be found here

Contributing

See the contributing guide for more information