The GitHub Actions job "Check newsfragment PR number" on 
airflow.git/fix/scheduler/dbdagbag-unbounded-cache has succeeded.
Run started by GitHub user jason810496 (triggered by jason810496).

Head commit for run:
4b308b23b5bc96fbdee46db6c24b4e9271a47094 / LIU ZHE YOU <[email protected]>
Bound the scheduler's deserialized Dag cache

The scheduler kept every Dag version it deserialized in a mapping that never
evicted, so a long-running scheduler grew with the number of versions it had
ever seen until it was restarted or OOM killed. Deployments that redeploy Dags
frequently accumulate versions fastest and hit this soonest.

A least-recently-used cap is the only thing that bounds this outright. An idle
timeout would not: the scheduler re-checks an entry on each lookup, which
re-arms its expiry, so a timeout reclaims a version only once its runs finish
and it stops being requested, leaving memory a function of the concurrently
active set rather than a fixed ceiling.

Deliberately not configurable here, so the fix stays small enough to
cherry-pick. Cache activity currently reports under the existing
api_server.dag_bag.* metrics; a scheduler-specific namespace, along with
configuration, follows separately.

closes: #69001

Report URL: https://github.com/apache/airflow/actions/runs/32218840844

With regards,
GitHub Actions via GitBox


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to