uranusjr commented on code in PR #71072:
URL: https://github.com/apache/airflow/pull/71072#discussion_r4439025600
##########
airflow-core/src/airflow/jobs/scheduler_job_runner.py:
##########
@@ -399,6 +423,14 @@ def __init__(
self._multi_team = conf.getboolean("core", "multi_team")
self._dag_tags_in_metrics = conf.getboolean("metrics",
"dag_tags_in_metrics", fallback=False)
self._max_partition_dag_runs_per_loop = MAX_PARTITION_DAG_RUNS_PER_LOOP
+ # Edge-triggers the "partition Dag run cap reached" audit Log row:
True once that row
+ # has been committed for the current backlog episode. Reset only when
an unlocked count
+ # shows the backlog at or below the cap: under `skip_locked` an empty
or short fetch
+ # can just mean other schedulers hold the rest.
+ # Process-local: each scheduler in an HA deployment writes at most one
audit row per
+ # episode; cross-process de-duplication would need a per-tick DB read,
defeating the
+ # point of edge-triggering.
+ self._partition_cap_backlog_reported = False
Review Comment:
The comment still says "the audit row" in sigular form, but this now writes
up to 20 rows. (Also in multiple other comments and log messages.)
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
