Replies: 1 comment
-
I think it could be the next step after implementing task-based SLAs. https://cwiki.apache.org/confluence/display/AIRFLOW/AIP-57+Refactor+SLA+Feature The effort to implement it is stalled a bit, but hopefully we can re-start it soon and this could be one of next steps. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Would be useful to implement an execution timeout that applies to all tasks in a task group cumulatively.
For example:
A DAG contains a Task Group with potentially 100s of mapped tasks. The user wants a timeout of 8hrs for the entire task group to complete. Due to concurrency and resource restraints, not all tasks in the group are executed concurrently, starting at different times. Setting
execution_timeout
of 8h on the mapped tasks themselves will not achieve the desired result, and it's easily possible for the task group to run for days, considering low concurrency, high task duration, and retries upon failures.If the "Overall Duration" concept that exists in the Web UI views could be extended into the scheduler in some way, then task group execution timeouts could be implemented.
Beta Was this translation helpful? Give feedback.
All reactions