Bjorn Olsen created AIRFLOW-6944:
------------------------------------
Summary: Allow AWS DataSync to "catch up" when Task is already
running
Key: AIRFLOW-6944
URL: https://issues.apache.org/jira/browse/AIRFLOW-6944
Project: Apache Airflow
Issue Type: Improvement
Components: aws
Affects Versions: 1.10.9
Reporter: Bjorn Olsen
Assignee: Bjorn Olsen
Current AWS DataSyncOperator attempts to start running a DataSync Task, with no
regard / check for whether the task is already running or not.
It is useful to have capability to optionally allow the operator to "catch up"
instead of starting the Task - if the Task is of a particular status eg
'QUEUED' then we might want to just use the currently Queued one instead of
submitting another one. This scenario can happen if the task was previously
submitted but the operator timed out waiting for it, for example.
Or, maybe we want to wait for the Queued task to complete and then submit
another Task ...
Allowing the user some options in terms of status management allows for various
robustness when submitting new task executions.
However, the current functionality of "Fail if the Task can't be started"
should remain default, to prevent unintentional problems which can arise if we
instead decided to always wait if there is already a task queued. For example
if the previous task has different Include filters than the new task, then
logically they aren't the same.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
