Mastering Local Development for GraphQL Federation - WunderGraph
Overview
In development, you aim to test your schema changes as quickly as possible, bypassing the need to check and update your federated graph on the control plane. This approach does not replace the necessity of pushing your subgraph to the control plane in production; however, it serves as a more expedient method for iterating on your graph during the development phase.
If you want to start your router in production with a static config please use the fetch command instead. This will fetch the latest valid production config from the control plane. wgc router compose should only be used for local development.
Prerequisites
Getting started
1. Add your subgraphs
In order to compose locally, we need to create a compose.yaml file that includes all the subgraphs you wish to include and compose into a federated graph. The information you are required to provide is as follows:
compose.yaml
version: 1
subgraphs:
- name: subgraph-a
routing_url: http://localhost:4001/graphql
# a) Specify a schema to introspect by file OR
schema:
file: ../schemas/subgraph-a.graphqls
- name: subgraph-b
routing_url: http://localhost:4002/graphql
# b) Specify introspection to introspect on a running subgraph
introspection:
url: http://localhost:4002/graphql
headers:
Authorization: 'Bearer YOUR_TOKEN_HERE'
| Property | Description | Required |
|---|---|---|
| name | The unique name of the subgraph | true |
| routing_url | The unique url (endpoint) of the subgraph (typically ends with /graphql) |
true |
| introspection.url | Required if you want to dynamically introspect a running subgraph server | false |
| introspection.headers | Headers to pass on the introspection request | false |
| schema.file | Path to the subgraph GraphQL schema | false |
2. Generate the Router config
After you have configured everything, you can generate the static router config as follows:
wgc router compose -i compose.yaml -o router.json
This command introspects all your subgraphs and produces a router.json that can be passed to the router in the next step.
3. Run the config with the router
Create a config.yaml file in the same directory as your router binary.
config.yaml
dev_mode: true
execution_config:
file:
# Path to the previous generated file
path: "router.json" # or EXECUTION_CONFIG_FILE_PATH
watch: true # EXECUTION_CONFIG_FILE_WATCH
graph:
# Result of `wgc router token create`. Can be omitted for local testing.
token: "" # GRAPH_API_TOKEN
We enabled the file watcher to hot-reload the server whenever you regenerate the router.json file. This is super handy for rapid-development.
If you omit the token, analytics and tracing are disabled. For production create a token wgc router token create and use polling instead. This ensures that the latest valid config is deployed to your routers automatically.
4. Open the playground
Finally, run the router and go to localhost:3002. You will see a playground and you’re ready to test your changes.
./router