Anurag-S1ngh opened a new pull request, #77087:
URL: https://github.com/apache/airflow/pull/77087

   ## Summary
   A trigger that fails and is still needed (for example an asset watcher whose 
system is down) was re-created about every 2 seconds for as long as it kept 
failing, each cycle writing its row and reopening connections. Re-creating is 
now delayed between attempts.
   
   ## Changes
   - The delay grows with the trigger's `consecutive_failures` and is capped by 
a configurable limit; the cap and the base delay are configurable (`[triggerer] 
failed_trigger_retry_delay`, default 5s; `[triggerer] 
failed_trigger_retry_max_delay`, default 300s)
   - Triggers brought over from another triggerer with a recorded failure count 
back off from the start instead of being re-created immediately
   - Watchers sharing a stream get the same consecutive failure count and the 
same delay, so they are re-created as one group
   - Deferred tasks' triggers still fail their task right away
   - Failure bookkeeping and triggerer logs are unchanged
   
   ## Testing
   - `uv run --project airflow-core pytest 
airflow-core/tests/unit/jobs/test_triggerer_job.py -q` (192 passed, 12 errors 
pre-existing on main)
   - `prek run ruff --files ...` and `prek run ruff-format --files ...` pass
   
   Closes #77052


-- 
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]

Reply via email to