Gabrielle Bennett
08/22/2024, 6:03 PMMarvin
08/22/2024, 6:04 PMtimeout_seconds
not enforced with RayTaskRunner
:
- Flows using the RayTaskRunner
may not enforce timeout_seconds
correctly. This means the flow can exceed the specified timeout without being interrupted.
- Link: https://github.com/PrefectHQ/prefect/issues/13004
2. Flows not transitioning to TimedOut
state immediately:
- There's an issue where a flow exceeding its timeout_seconds
parameter remains marked as Running
instead of transitioning to the TimedOut
state immediately.
- Link: https://github.com/PrefectHQ/prefect/issues/7645
The documentation also highlights that flow timeouts are supposed to fail the flow and mark it visibly as TimedOut
. Link to Docs: https://docs.prefect.io/latest/concepts/flows/
Given this, your issue might be related to one of these bugs. You can follow the links for more details and potential fixes.Gabrielle Bennett
08/22/2024, 6:06 PMMarvin
08/22/2024, 6:06 PMDbtCoreOperation
to run a DBT model where individual nodes run longer than 1800 seconds, the entire run command is aborted with a Query cancelled on user's request
message.
- Details: The issue happens when any individual node exceeds 1800 seconds of runtime. There doesn't seem to be an explicit timeout parameter to control this behavior.
- Link: https://github.com/PrefectHQ/prefect/issues/12954
This might be related to your timeout issue with DBT processes. You can follow the issue for updates or potential workarounds.