If there are no flow logs at all, the best places to look are usually outside the flow itself: 1. The flow run page in Prefect UI - Open the run and check its state message - If the process crashed before your flow could initialize logging, the useful clue is often here rather than in the Logs tab - Look for messages like: -

Flow run infrastructure exited with non-zero status code ...

-

Failed to submit flow run to infrastructure ...

-

Lost connection to flow run infrastructure ...

2. Your worker logs - This is often the most important place - The worker usually records why the flow process/container failed, even if the flow never emitted logs - Common clues: - bad entrypoint / command not found - dependency/import problems very early in startup - OOM / SIGKILL - infrastructure submission failures 3. Infrastructure logs Depending on how you run the flow: - Process worker: check the shell/session where the worker is running - Docker worker:

docker logs <container-id>

- Kubernetes worker:

kubectl logs <pod-name>

and

kubectl describe pod <pod-name>

- ECS: CloudWatch logs / task exit status 4. Exit codes If you see an exit code in the state message or worker logs, that’s a strong clue: -

127

= command not found / bad entrypoint -

126

= permission problem -

137

or

-9

= often OOM kill / SIGKILL -

143

or

-15

= SIGTERM / graceful termination -

1

= general runtime failure Rule of thumb - If the run has logs: the flow code probably started, so inspect the flow logs - If the run has no logs: the process likely failed before Prefect logging was fully set up, so inspect the worker + infrastructure logs Useful docs: - Workers - Work pools and infrastructure If you want, tell me what worker type you’re using (

process

,

docker

,

kubernetes

,

ecs

, etc.) and what state message you see in the UI, and I can tell you exactly where to look next.