Re: [LAZY CONSENSUS] backfill.max_active_runs distinct from DAG.max_active_runs

2024-10-15 Thread Daniel Standish
Belated confirmation that this vote has passed. Will update the docs.

Re: [LAZY CONSENSUS] backfill.max_active_runs distinct from DAG.max_active_runs

2024-10-07 Thread Jarek Potiuk
Sounds reasonable. No objection. On Mon, Oct 7, 2024 at 1:47 PM Daniel Standish wrote: > That's right. Completely separate. > > Backfill max active runs will be a setting per-backill job, each of which > can contain many dag runs of course. > > Neither count will affect the other, and neither w

Re: [LAZY CONSENSUS] backfill.max_active_runs distinct from DAG.max_active_runs

2024-10-07 Thread Daniel Standish
That's right. Completely separate. Backfill max active runs will be a setting per-backill job, each of which can contain many dag runs of course. Neither count will affect the other, and neither will be affected by the number of running dag runs of the other class (i.e. backfill vs non-backfill)

Re: [LAZY CONSENSUS] backfill.max_active_runs distinct from DAG.max_active_runs

2024-10-07 Thread Jens Scheffler
Hi Daniel, No objection but just a (late arrival) question I had... The `Backfill.max_active_runs` Will be measured, limited and counted irrespective of the `DAG.max_active_runs` and irrespective of the real actual load? Assuming `DAG.max_active_runs == 10` and currently 5 DAGs are running from