o-nikolas opened a new pull request, #54523:
URL: https://github.com/apache/airflow/pull/54523
### Summary
LocalExecutor would only actually begin execution for every second task
submitted to it. This change fixes this behaviour.
The number of queued tasks would need to exceed the number of currently
running tasks for a new task to be executed, which is logically close to what
is needed but not quite. This means if 4 tasks are running and 4 are pending
none of the 4 tasks will begin, but if a 5th task is queued it will begin and
there will be 5 tasks running and 4 waiting, it will take another two tasks
submitted (for a total of 5 then 6 pending tasks) before the next is started,
and so one. So in reality one of every two tasks is started.
The logic is really as simple as: if we have any pending tasks and we are
still within our parallelism limits, start those tasks running. This schedules
tasks in accordance to user expectations and how Airflow 2.X scheduled tasks
for the LocalExecutor (since this regressed during the migration from 2.X to
3.X for the LocalExecutor)
### QA:
Testing a simple dag that simply starts many parallel tasks. It should start
as many concurrently as the default configuration allows, the default limit of
max active tasks per dag of 16 should be the limit hit:
```
from datetime import datetime, timedelta
from airflow import DAG
from time import sleep
from airflow.decorators import task
import socket
import random
default_args = {
'owner': 'Airflow',
'depends_on_past': False,
'start_date': datetime(2023, 1, 1),
'email_on_failure': False,
'email_on_retry': False,
'retries': 1,
'retry_delay': timedelta(minutes=1),
}
max_tasks = 25
with DAG(
'slow_task_dag',
default_args=default_args,
schedule=None, # Changed from schedule_interval to schedule
catchup=False
) as dag:
for y in range(0, max_tasks):
@task(task_id=f'print_hello_{y}')
def task_method(ds=None, **kwargs):
print(f'Getting executed at worker {socket.gethostname()}')
sleep(10 + random.randint(0, 9))
task_method()
```
Before this change you would see this execution pattern where only 8 tasks
would actually start running (incorrectly) despite 16 being queued (correctly).
<img width="1170" height="788" alt="Screenshot from 2025-08-13 10-46-53"
src="https://github.com/user-attachments/assets/d522ab3c-121b-4499-9af8-bc0c2e01178d";
/>
With this change all 16 queued tasks begin running:
<img width="1129" height="771" alt="Screenshot from 2025-08-14 12-48-04"
src="https://github.com/user-attachments/assets/08a5f05e-d786-4873-8a87-2b6c88d9c518";
/>
<!--
Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements. See the NOTICE file
distributed with this work for additional information
regarding copyright ownership. The ASF licenses this file
to you under the Apache License, Version 2.0 (the
"License"); you may not use this file except in compliance
with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing,
software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
KIND, either express or implied. See the License for the
specific language governing permissions and limitations
under the License.
-->
<!--
Thank you for contributing! Please make sure that your code changes
are covered with tests. And in case of new features or big changes
remember to adjust the documentation.
Feel free to ping committers for the review!
In case of an existing issue, reference it using one of the following:
closes: #ISSUE
related: #ISSUE
How to write a good git commit message:
http://chris.beams.io/posts/git-commit/
-->
<!-- Please keep an empty line above the dashes. -->
---
**^ Add meaningful description above**
Read the **[Pull Request
Guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#pull-request-guidelines)**
for more information.
In case of fundamental code changes, an Airflow Improvement Proposal
([AIP](https://cwiki.apache.org/confluence/display/AIRFLOW/Airflow+Improvement+Proposals))
is needed.
In case of a new dependency, check compliance with the [ASF 3rd Party
License Policy](https://www.apache.org/legal/resolved.html#category-x).
In case of backwards incompatible changes please leave a note in a
newsfragment file, named `{pr_number}.significant.rst` or
`{issue_number}.significant.rst`, in
[airflow-core/newsfragments](https://github.com/apache/airflow/tree/main/airflow-core/newsfragments).
--
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]
