GraphQL Summit is back for three days of insights, hands-on learning, and fun to celebrate the GraphQL community. Join us in San Diego Oct 3-5.
Docs
Try Apollo Studio

Health checks

Determining the health status of the Apollo Server


Health checks are often used by load balancers to determine if a server is available and ready to start serving traffic.

GraphQL-level health checks

The most natural way to determine if your GraphQL server is healthy is to run a GraphQL operation!

Every GraphQL server supports the trivial query that asks for the "type name" of the top-level query. So you can run a health check as a GET against an URL like:

https://your.server/graphql?query=%7B__typename%7D

This health check will run a real GraphQL operation, so if your server requires special headers or cookies in order to run any query, you'll need to provide that as well.

HTTP-level health checks

Apollo Server also supports a simple HTTP-level health check. This is enabled by default and is served at the URL path /.well-known/apollo/server-health. This returns the 200 status code if the HTTP server is successfully serving. It does not invoke any of the GraphQL execution machinery. This means it may be better for your needs if your server's context function requires you to be logged in, but it may be worse for your needs if you want your health check to test that your server's context function runs successfully.

In the batteries-included apollo-server package

The batteries-included apollo-server package doesn't provide you with direct access to your HTTP server, so we provide a few knobs to let you adjust the HTTP-level health check.

You can pass a string healthCheckPath to the ApolloServer constructor to change the URL where the health check is served. You can also pass healthCheckPath: null to disable the health check.

If you'd like the health check to do more than just "always return success", you can pass an async function onHealthCheck function to the ApolloServer constructor. If defined, this onHealthCheck async function should return if the server is deemed ready or throw if there is an error. Returning (resolving the Promise) will result in an HTTP status code of 200, which is generally desired by most health-check tooling (e.g. Kubernetes, AWS, etc.), while throwing (rejecting the Promise) will result in an HTTP status code of 503.

import { ApolloServer, gql } from 'apollo-server';
import { ApolloServerPluginLandingPageLocalDefault } from 'apollo-server-core';
// Undefined for brevity.
const typeDefs = gql``;
const resolvers = {};
const server = new ApolloServer({
typeDefs,
resolvers,
csrfPrevention: true,
cache: 'bounded',
async onHealthCheck() {
if (everythingLooksHealthy()) {
return;
} else {
throw new Error('...');
}
},
plugins: [
ApolloServerPluginLandingPageLocalDefault({ embed: true }),
],
});
server.listen().then(({ url }) => {
console.log(`🚀 Server ready at ${url}`);
console.log(
`Try your health check at: ${url}.well-known/apollo/server-health`,
);
});

Note that the healthCheckPath and onHealthCheck options are only accepted by the batteries-include apollo-server constructor, not by other framework integration packages.

In framework integration packages

Like apollo-server, framework integration packages like apollo-server-express and apollo-server-lambda also serve a health check at the URL path /.well-known/apollo/server-health by default. Customizing this health check is done in the framework-specific middleware function such as applyMiddleware or createHandler rather than in the ApolloServer constructor.

To disable serving the health check endpoint, pass disableHealthCheck: true to the framework-specific middleware function.

These packages also support an onHealthCheck function like apollo-server; this function is passed to the framework-specific middleware function rather than to the ApolloServer constructor. If you're using apollo-server-azure-functions (which doesn't provide a general way to customize its HTTP serving behavior), this option may be helpful. For other integrations such as apollo-server-express, we don't recommend that you customize the health check via Apollo. Since you're already setting up your web framework, if you need to customize the behavior of the health check then it is probably more straightforward to simple define a health check handler yourself directly in your web framework (and pass disableHealthCheck: true to disable Apollo Server's health check). A future major version of Apollo Server may change the health check feature to be specific to the "batteries-included" apollo-server package rather than part of all framework integration packages. For similar reasons, you can't customize the health check path in framework integration packages; just disable the built-in health check and add your own at your preferred path.

Edit on GitHub
Previous
Metrics and logging
Next
ApolloServer