Hey! We've encountered a bug in the Kubernetes work pool. When running a flow that has a task with a task run concurrency limit, if that flow receives an OOM kill from Kubernetes, it doesn't report anything to the Prefect server. Which means that the flow remains in a running state, and its task run concurrency limit stays locked which prevents my other tasks from starting. I would love Prefect to somehow detect these kind of crashes (maybe a sidecar / watchdog?) and report them back. Thanks!

We've seen something similar on ECS Fargate (we think. Currently investigating.) Interested to hear what Prefect has to say.

@Alexander Azzam @Nate Any insight on this?

Hey @Bil Tal, did you ever resolve this? We're having the same issue with tag based concurrency