[ecs-fargate] [request]: improved/detailed task event log

[dekimsey]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment

Tell us about your request Improve the task-specific event logging information. (Information from the ECS daemon itself about it's actions, not related to container logging.) I would love to see logged information about task-specific actions the daemon is performing at a high-level.

Which service(s) is this request for? ECS and in particular Fargate

Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard? Debugging multiple-container start-up is very tricky right now. ECS Fargate will:

• Trim Stopped Reason to ~ 128 characters, making long ARN-containing messages useless.
• Emits no information regarding when healtchecks are executing, or their status.
• Discards healthcheck statuses even after reaching success when the task is killed.
• Marks healthchecks statuses (UNKNOWN) on task failures, making diagnosing the failure after it has stopped nigh-impossible.
• Fails to indicate which task failed the healthcheck that caused the entire task to be aborted.
• No perceptible error messages used when the incorrect image architecture is used.
• Zero timestamped information related to any of the above.

These are just the items that have occurred to me in the last few minutes. I'd be comfortable stating debugging ECS Fargate containers in general is currently quite difficult. Especially after the fact (since there is no information saved, nor is any of it timestamped if it is available).

Generally speaking the, inability to see even a high-level event log of what the daemon was doing and when for a given task is a large hindrance to debugging start-up issues. In the past, I've reached out to support which can take days, and their response is usually did you try running the task in ECS EC2? Which isn't very helpful as they aren't the same service and behave differently.

Are you currently working around this issue? Trial and error, and a great deal of patience.

Additional context None

Attachments None