Meant to add more... This is happening in a docker container and the port number varies.

I can see my

node

process running, and a prisma query engine process start sometimes. I thought there was an issue with connection pooling, so I setup pgbouncer and connect to that.

Hmm... I just bumped up the memory on my container, and it looks like it's more stable now

fwiw... We're using

OpenFaaS

on kubernetes and these are hosted as

OpenFaaS

"microservice" functions, so the process stays up between requests. The

graphql

app was original poc using

prisma

and we're splitting the

primsa

schema into smaller, self-contained schemas for the microservices they represent. The

graphql

server will eventually only make calls to the other microservices and won't have

prisma

Soon, we'll probably get more into optimization and trying to determine the best strategy for scaling the microservices, but for now, we're just increasing memory.