A starting point for deploying the Router via a Dockerfile using GraphOS Enterprise.

⚠️This template requires GraphOS Enterprise, if you'd like a similar template that does not require an enterprise plan, please open a discussion and describe your use-case and requirements.

What's included

• router.yaml—configuration for the router
• Dockerfile—used to build the router for deployment
• .apollo—contains some JSON schemas for the config files (to make IDE experience better)
• .github/workflows/update-router-config.yaml—auto-update JSON schemas when the router version changes
• .vscode—contains recommended VS Code settings
• .idea—contains recommended Jetbrains editor settings
• renovate.json—configured to keep Router up to date
• A sample supergraph.yaml file for testing Router via rover dev. You'll need to update this file to point at the local versions of your subgraphs.

Next steps

• Deploy to your environment of choice and set up CI/CD to deploy newer versions.
  • Set APOLLO_KEY and APOLLO_GRAPH_REF secrets in deploy
  • (Optional) - Delete irrelevant cloud provider config files (i.e. if deploying to Railway, you can delete the render.yaml file)
• Enable Renovate on this repo to keep Router up to date
• Set up a deployment preview for PRs to changes which you can run integration tests against
• Set up secrets for each of your subgraphs so that only the routers can access them
• Disable subgraph error inclusion, sandbox, and introspection in router.yaml once you have everything working (you probably don't want those enabled in production)

Commands

• docker build -t router . builds the router image with the tag router for local testing.
• rover dev --supergraph-config supergraph.yaml --router-config router.yaml to run the Router locally without Docker (using Rover). You'll need to update the supergraph.yaml file to point at the local versions of your subgraphs. Make sure to set the required environment variables ahead of time!
  • You can add the --graph-ref option to your rover dev command if you want to develop off of your current variant's state in GraphOS
• docker run -it --env APOLLO_KEY --env APOLLO_GRAPH_REF -p4000:4000 router runs the same router image you'll run in production. You can now query the router at http://localhost:4000.
  • Make sure to set the env vars APOLLO_KEY and APOLLO_GRAPH_REF first
  • You can alternatively create a file (e.g., .env) and run docker run -it --env-file .env -p4000:4000 router. Make sure not to check this file into source control!
  • Your local router will need network access to the subgraphs