v5 (latest)FeaturesApollo Federation

Apollo Federation

Apollo Federation is a specification that applies microservice architecture through GraphQL APIs.

If you’re new to GraphQL Federation, learn how it combines your GraphQL APIs into a unified supergraph.

Gateway

You can use GraphQL Yoga to implement a Gateway exposing an Apollo Federation supergraph. For this, the gateway needs to have access to the supergraph SDL.

You have the choice between loading it from a schema registry (like Hive or Apollo GraphOS) or locally from the file system.

Instead of using GraphQL Yoga to implement the gateway, we recommend to use Hive Gateway, our open-source MIT-licensed GraphQL gateway for Apollo Federation.

Using Hive

Hive is a schema registry and usage reporting plateform for GraphQL. You can use it to manage an Apollo Federation supergraph.

For a full step by step setup getting started, please follow this Apollo Federation usage guide.

Installation

npm i graphql-yoga graphql @graphql-hive/yoga

Example

src/index.mjs
import { createServer } from 'node:http'
import { createYoga } from 'graphql-yoga'
import { createSupergraphSDLFetcher } from '@graphql-hive/yoga'
import { getStitchedSchemaFromSupergraphSdl } from '@graphql-tools/federation'
 
const supergraphFetcher = createSupergraphSDLFetcher({
  key: env.HIVE_CDN_KEY,
  endpoint: env.HIVE_CDN_URL
})
const { supergraphSdl } = await supergraphFetcher()
 
const yoga = createYoga({
  schema: getStitchedSchemaFromSupergraphSdl({ supergraphSdl })
})
 
const server = createServer(yoga)
 
server.listen(4000, () => {
  console.info('Server is running on http://localhost:4000/graphql')
})

You can also check the Hive plugin documentation for more information

Configuration

A full configuration guide can be found in Hive’s “Configuration” page.

Using Apollo Managed Federation

Apollo GraphOS can be used as a schema registry when using Managed Federation.

You can use GraphQL Yoga to implement a Gateway exposing the Managed Federation supergraph stored in GraphOS by using @graphql-yoga/apollo-managed-federation. This plugin will automatically fetch the supergraph SDL from GraphOS, sticth it into an executable schema and serve it.

It also starts a polling mechanism to keep the schema up to date in the background and handle retry on failure.

Installation

npm i graphql-yoga graphql @graphql-yoga/apollo-managed-federation

Example

src/index.mjs
import { createServer } from 'node:http'
import { createYoga } from 'graphql-yoga'
import { useManagedFederation } from '@graphql-yoga/apollo-managed-federation'
 
const yoga = createYoga({
  plugins: [useManagedFederation()]
})
 
const server = createServer(yoga)
 
server.listen(4000, () => {
  console.log('Server is running on http://localhost:4000')
})
 
process.on('SIGINT', () => {
  server.close()
})

You can also check a full example on our GitHub repository here

Configuration

By default, the plugin will use standard Apollo GraphOS environment variables.

APOLLO_KEY='<YOU_APOLLO_API_KEY>' APOLLO_GRAPH_REF='<YOUR_GRAPH_ID>[@<VARIANT>]' node src/index.mjs

If you don’t know where to find these values, please refer to the Apollo GraphOS documentation.

You can also provide this configuration programaitically by providing a configurtion object to the plugin, along with other options to customize the polling/retry behavior.

const yoga = createYoga({
  plugins: [
    useManagedFederation({
      apiKey: '<YOU_APOLLO_API_KEY>',
      graphRef: '<YOUR_GRAPH_ID>[@<VARIANT>]',
      maxRetries: 5, // max retries in case of loading failure
      retryDelaySeconds: 0, // delay between retries
      minDelaySeconds: 1 // minimum delay between polling requests
    })
  ]
})

Other more advanced options are available, please refer to the type definitions for more information.

Using the file system

To create a gateway server without using Apollo GraphOS, you can use the @graphql-tools/federation which can generate an executable schema from a supergraph SDL.

Installation

npm i graphql-yoga graphql @graphql-tools/federation

Example

To generate the schema, we first need the supergraph SDL. This can be generated by the Apollo Rover CLI or fetched from a schema registry (like Hive or Apollo GraphOS).

To generate the supergraph SDL locally, you first have to define a supergraph configuration by creating a supergraph.yaml file.

supergraph.yaml
subgraphs:
  accounts:
    routing_url: http://localhost:4001/graphql
    schema:
      file: ./services/accounts.graphql
  inventory:
    routing_url: http://localhost:4002/graphql
    schema:
      file: ./services/inventory.graphql

You can then use the Apollo Rover CLI to generate the supergraph SDL.

rover supergraph compose --config ./supergraph.yaml > supergraph.graphql

You can now serve the gateway using GraphQL Yoga.

src/index.mjs
import { readFileSync } from 'fs'
import { createServer } from 'node:http'
import { createYoga } from 'graphql-yoga'
import { getStitchedSchemaFromSupergraphSdl } from '@graphql-tools/federation'
 
const yoga = createYoga({
  schema: getStitchedSchemaFromSupergraphSdl({
    // This doesn't have to be from a file system, it can be fetched via HTTP from a schema registry
    supergraphSdl: readFileSync('./supergraph.graphql', 'utf-8')
  })
})
 
const server = createServer(yoga)
 
server.listen(4000, () => {
  console.log(`🚀 Server ready at http://localhost:4000/graphql`)
})

Handling Subgraph Errors

By default, GraphQL Yoga masks any unexpected GraphQL Errors. This is done to prevent leaking internal errors to the client. If you know that your subgraph is safe and you want to expose the errors to the client, you can customize the error masking bahviour.

Learn more about error masking

Expose subgraph errors to the client
import { createYoga, maskError } from 'graphql-yoga'
import { schema } from './schema.js'
 
const yoga = createYoga({
  schema,
  maskedErrors: {
    maskError(error, message, isDev) {
      if (error?.extensions?.code === 'DOWNSTREAM_SERVICE_ERROR') {
        return error
      }
 
      return maskError(error, message, isDev)
    }
  }
})

Federation Service

You don’t need any extra plugins for Yoga for Federation Service.

Installation

npm i graphql-yoga @apollo/subgraph graphql

Example User Service

const { parse } = require('graphql')
const { buildSubgraphSchema } = require('@apollo/subgraph')
const { createYoga } = require('graphql-yoga')
const { createServer } = require('http')
 
const typeDefs = parse(/* GraphQL */ `
  type Query {
    me: User
  }
 
  type User @key(fields: "id") {
    id: ID!
    username: String
  }
`)
 
const resolvers = {
  Query: {
    me() {
      return { id: '1', username: '@ava' }
    }
  },
  User: {
    __resolveReference(user, { fetchUserById }) {
      return fetchUserById(user.id)
    }
  }
}
const yoga = createYoga({
  schema: buildSubgraphSchema([{ typeDefs, resolvers }])
})
 
const server = createServer(yoga)
 
server.listen(4001, () => {
  console.log(`🚀 Server ready at http://localhost:4001`)
})

Federated tracing

Inject additional metrics for Apollo’s federated tracing.

You’ll need the @graphql-yoga/plugin-apollo-inline-trace Yoga plugin for this.

Installation

npm i graphql-yoga @graphql-yoga/plugin-apollo-inline-trace graphql

Example Federated tracing

import { createServer } from 'http'
import { createYoga } from 'graphql-yoga'
import { useApolloInlineTrace } from '@graphql-yoga/plugin-apollo-inline-trace'
 
const yoga = createYoga({
  plugins: [
    useApolloInlineTrace()
    // ...rest of your Apollo federation plugins
  ]
})
 
// Start the server and explore http://localhost:4000/graphql
const server = createServer(yoga)
server.listen(4000, () => {
  console.info('Server is running on http://localhost:4000/graphql')
})

Working Example

Check our working example to try it out.

CookiesTesting