Fix bug in LivyOperator when its trigger times out #38916
Merged
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Related issue
closes: #37898
Explain
When a LivyOperator was instantiated with deferrable=True and its batch job ran for more time than the set execution_timeout, airflow would detect this timeout and would cancel the trigger and then try to kill the task with the 'on_kill' method.
But that would fail raising an AttributeError because the batch_id attribute wouldn't be defined in the constructor method.
From now on, the LivyTrigger will timeout itself before airflow does it, and it will send an event to the LivyOperator signaling that a timeout happened. This way, the operator can stop the running Livy batch job, and can fail the task instance gracefully.