mis98zb commented on code in PR #32122:
URL: https://github.com/apache/airflow/pull/32122#discussion_r1264922227


##########
airflow/jobs/scheduler_job_runner.py:
##########
@@ -277,9 +288,51 @@ def __get_concurrency_maps(self, states: 
Iterable[TaskInstanceState], session: S
             .where(TI.state.in_(states))
             .group_by(TI.task_id, TI.run_id, TI.dag_id)
         )
-        return ConcurrencyMap.from_concurrency_map(
-            {(dag_id, run_id, task_id): count for task_id, run_id, dag_id, 
count in ti_concurrency_query}
+        ti_concurrency = {
+            (dag_id, run_id, task_id): count for task_id, run_id, dag_id, 
count in ti_concurrency_query
+        }
+
+        tg_concurrency_query: Iterator[tuple[str, str, str, int]] = 
session.execute(
+            select(TI.task_id, TI.run_id, TI.dag_id, TI.map_index)
+            .where(
+                TI.map_index >= 0,
+                TI.state.in_(
+                    (
+                        TaskInstanceState.SCHEDULED,
+                        TaskInstanceState.QUEUED,
+                        TaskInstanceState.RUNNING,
+                        TaskInstanceState.UP_FOR_RESCHEDULE,
+                        TaskInstanceState.UP_FOR_RETRY,
+                    )

Review Comment:
   SHUTDOWN will go FAILED, which is deprecated.
   RESTARTING / DEFERRED will be retried.
   These three are fine for concurrency limitation.
   
   However, None cannot be used for identify active task group.
   
   So we cannot use `State.unfinished`.
   
   Do we need to add a state list like `State.active`? @uranusjr 



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscr...@airflow.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org

Reply via email to