shivaam opened a new pull request, #70455:
URL: https://github.com/apache/airflow/pull/70455

   CeleryExecutor currently publishes each task to Redis in a separate network
   round trip. This makes publication time grow with both the number of 
scheduled
   tasks and broker latency.
   
   This change lets Celery and Kombu prepare each task message normally, then
   buffers the final Redis queue writes in a non-transactional pipeline and
   flushes them together. Callback workloads and non-Redis brokers keep the
   existing publication path.
   
   Redis pipelining is enabled by default and can be disabled with
   `[celery] redis_pipelined_publish_enabled = False`. The implementation 
mirrors
   Kombu's private Redis transport methods, so a focused contract test will fail
   if a future Kombu update changes that integration point.
   
   Local validation:
   
   - Celery executor unit module: 74 passed, 6 skipped.
   - Linux multiprocessing regression: passed 250 repeated dispatch cycles.
   - PostgreSQL/Celery integration suite: 10 passed with real Redis and RabbitMQ
     brokers and Celery workers.
   - Real daemon load: scheduler, API server, DAG processor, triggerer, 
PostgreSQL,
     Redis, and a four-process Celery worker completed 1,000/1,000 Bash tasks 
and
     four/four DAG runs successfully. Redis recorded exactly 1,000 queue writes
     and drained to zero.
   - Changed source files pass `mypy`; regular `prek` checks pass. The
     provider-wide manual `mypy` hook was terminated by the local Docker 
resource
     limit (exit 137/143).
   
   closes: #8854
   
   ---
   
   ##### Was generative AI tooling used to co-author this PR?
   
   - [X] Yes — Codex (GPT-5)
   
   Generated-by: Codex (GPT-5) following [the 
guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions)
   
   ---
   Drafted-by: Codex (GPT-5) (no human review before posting)
   


-- 
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