High memory usage leading to OOM killed on kubernetes

[farru1998]

Summary

Hey folks!

Dealing with an issue that the memory usage of NextJS app on kubernetes is increasing linearly over time.
As per our observation, it seems the garbage collector is not being triggered maybe. As a result, memory usage increases linearly, eventually reaching the pod’s memory limit and causing an OOM (Out of Memory) error.

Once this happens, the Kubernetes pod is restarted. Below is the attached memory trend before and after restart.

We are running npm start in our dockerfile as the CMD command which eventually runs:

start": "next start -p 5000"

Additional information

Next Version: 14.2.5
Node Version: 20

Example

No response

[icyJoseph]

IIRC, we have to specify max-old-space-size? https://stackoverflow.com/a/48392705