Minsu Jo
03/13/2025, 2:24 PMjobmanager.scheduler: adaptive
jobmanager.adaptive-scheduler.min-parallelism-increase: 1
job.autoscaler.enabled: "true"
job.autoscaler.stabilization.interval: 20s
job.autoscaler.metrics.window: 5m
job.autoscaler.target.utilization: "0.2"
job.autoscaler.target.utilization.boundary: "0.1"
job.autoscaler.restart.time: 2m
job.autoscaler.catch-up.duration: 5m
I have observed that the autoscaler is functioning as expected.
These are the logs related to scaler.
...
2025-03-13 00:16:10,662 INFO org.apache.flink.runtime.resourcemanager.active.ActiveResourceManager [] - need request 4 new workers, current worker number 5, declared worker number 9
...
2025-03-13 01:27:41,643 INFO org.apache.flink.runtime.resourcemanager.active.ActiveResourceManager [] - need release 1 workers, current worker number 9, declared worker number 8
...
However, I am curious about what exactly triggers the scaling process. Could you clarify which factors influence the scaling decisions?