-
Notifications
You must be signed in to change notification settings - Fork 14k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
max_active_runs doesn't seem to work anymore (in 1.10.14) #13802
Labels
kind:bug
This is a clearly a bug
Comments
As far as I can tell: https://github.com/apache/airflow/blob/1.10.14/airflow/jobs/scheduler_job.py#L794 never actually populates "active_dag_runs" anymore like in 1.10.12 https://github.com/apache/airflow/blob/1.10.12/airflow/jobs/scheduler_job.py#L796 |
Looks like it broke here: cb8d53f |
closed by #13803 |
abhishekbafna
pushed a commit
to twitter-forks/airflow
that referenced
this issue
Aug 25, 2021
…runs * EWT-797: Fixing max_active_runs to control the number of runs running simultaneously * Bumped the version Co-authored-by: Shruti Mantri <smantri@twitter.com> I referred to the git issue: apache#9975 which in its first line itself tells that the issue that happens related to max_active_runs in 1.10.14 is a separate one and is discussed here: apache#13802 This issue was resolved but the fix went in the 1.10.15 version. I took the same fix: https://github.com/apache/airflow/pull/13803/files and have applied it in this commit (5e87232). The resolution worked and I did the testing on gke devel cluster: https://airflow-smantri--devel--etl-workflow-users.service.qus1.twitter.biz/admin/airflow/tree?dag_id=bq_dal_integration_dag Prior to the fix, if the max_active_runs = 1, and if you start more than 1 run simultaneously, all of them will start their tasks, and would not honour max_active_runs configuration. With this fix, if the max_active_runs = 1, and one of the runs is going on while the second run is triggered, the second run starts with the start_date being the time when it is triggered, but the second run won't start any of its tasks until the prior run is complete.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Apache Airflow version:
1.10.14
Environment:
Regular Airflow setup:
Python 3.6.2
Airflow 1.10.14
Celery 4.3.0
What happened:
max_active_runs isn't limiting DAG runs on DAGs when None schedule_interval (unsure about with a schedule, I don't use them). I set it to 20 but more than 20 are running.
It used to work on older versions of 1.10.
What you expected to happen:
DAG Runs to be limited to what I set in max_active_runs
Anything else we need to know:
I've had "max_active_runs" set as a DAG parameter for a long time and it always worked fine until this latest update I went from 1.10.12 -> 1.10.14.
I didn't see anything relevant in the upgrade changelogs, so I didn't realize it wasn't working until I look into some problems.
More so, the 0/20 doesn't seem to display in the RBAC UI, but that could be a change in the new UI since I just switched to that as well.
The text was updated successfully, but these errors were encountered: