Demonstrating Apollo issue #3052

This is a barebones example that demonstrates the apollo-server issue #3052.

Symptoms

When using a managed federated gateway, the underlying service errors are not properly relayed through to the client.

Steps

1. Create a new apollo engine project and setup your .env file with the ENGINE_API_KEY.
2. Initial Run
   • Comment out line 4 of src/gateway.js
   • Uncomment lines 6-20 of src/gateway.js
3. Execute npm start to begin the services
4. In a new terminal run the following commands to push managed services
   • npm run push:serviceA
   • npm run push:serviceB
5. For shits and giggles, navigate to http://localhost:4000 and execute the following query to see that the error is relayed properly.

   query Ping {
     pingServiceA
     pingServiceB
   }

6. Revert comments from above
   • Uncomment line 4 of src/gateway.js
   • Comment out lines 6-20 of src/gateway.js
7. Execute npm start
8. Follow step 5 and you will see an error that does not contain the message that is expected, but rather something alone the lines of: "addProtobufError called after stopTiming!"