Hi all, I'm having issues with flows that run on ECS via Fargate initially showing up with "Crashed" state. Each time after being in "Pending" state for 2 minutes. This "Crashed" state persists for about a minute, after which it switches to "Running". Eventually, they get completed without issues. I have tried to adjust several timeout settings for our prefect-server deployment but the issues remains. We are using Prefect v2.20.16. I saw that a while ago
@jack had a similar issue:
https://linen.prefect.io/t/23173237/we-are-having-issues-with-flows-that-run-on-ecs-fargate-init
I suspect that the flows enter the crashed state due to Fargate's cold-start taking too long, and Prefect interprets this as an infrastructure failure. Has anyone encountered this/ found a fix for this? Any suggestions are appreciated!