SIGKILL When running next start in ECS

[gustavotrapp]

I'm using NextJS 14 and Prisma, we have a multi tenanted database and are experiencing something really weird that started happening after we deployed our NextJS app to production, we have an ECS with 0.5 vCPU and 1 GB mem.

We get a lot of too many instances warning from prisma and think this is somehow causing a memory leak that sigkill the app.

It usually happens when our app has the most amount of traffic so we are thinking it might something to do with that, there are too many users logged in and the prisma instances just spike and it sigkills the app.

Here is the log of it happening, this will happen anywhere from twice to up to 5 times a day