MaxListenersExceededWarning: prisma 2.20.x and above

Summary

When upgrading @prisma/client and prisma from 2.19.0 to 2.20.x or newer, including 2.21.x, six in total “MaxListenersExceededWarning” messages are being logged:

(node:4511) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 SIGINT listeners added to [process]. Use emitter.setMaxListeners() to increase limit

(node:4511) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 SIGTERM listeners added to [process]. Use emitter.setMaxListeners() to increase limit

(node:4511) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 exit listeners added to [process]. Use emitter.setMaxListeners() to increase limit

(node:4511) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 SIGUSR2 listeners added to [process]. Use emitter.setMaxListeners() to increase limit

(node:4511) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 beforeExit listeners added to [process]. Use emitter.setMaxListeners() to increase limit

(node:4511) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 SIGUSR1 listeners added to [process]. Use emitter.setMaxListeners() to increase limit

After the initial logs they don’t reappear until re-starting our GraphQL server. Prior to and with versions 2.19.0 this did not happen.

Background

We run our GraphQL Server as a serverless function on Lambda and create it with the apollo-server-lambda. We are seeing these logs currently in our local development environment where we run the Lambda locally via the serverless-offline plugin as we utilize the serverless framework for our deployment.

Dependency Versions

• @prisma/client: 2.20.0 - 2.21.1

• nexus-plugin-prisma: 0.34.1

• prisma (dev): 2.20.0 - 2.21.1

• serverless-offline: 6.6.0

• apollo-server-lambda: ^2.19.0

Problem is seen using Node versions 12.19.1 and as of submitting latest 12 release 12.22.1

[shane-abcya]

This should be fixed in 2.21.2 let us know if it’s the case, if not, we can reopen this issue.

I can confirm the max listener warnings are gone. Thank you so much

[dborstelmann]

@Jolg42 Just fixed it in deployments. It was context.callbackWaitsForEmptyEventLoop = false;. For some reason the latency is still there in offline. I sent @ryands17 a replication repo, I’ll post a new issue if we discover something on the prisma side. Thanks!