SEZ9 commented on issue #11663: URL: https://github.com/apache/seatunnel/issues/11663#issuecomment-5203022967
This is a significant architectural change; please raise it on the dev mailing list and consider drafting a SEIP/design doc covering the control loop, metric inputs, decision policy, and actuation mechanism before implementation starts. Could you clarify the actuation scope: will the autoscaler only emit recommendations (metrics/REST) that external systems like Kubernetes HPA or an operator act on, or will it directly provision/decommission workers? Direct provisioning outside Kubernetes needs a pluggable executor design. Scale-in safety needs special attention in the design: how running jobs/pipelines on a to-be-removed worker are drained or rescheduled (checkpoint-based restore vs. graceful task migration), plus cooldown/hysteresis defaults to avoid oscillation. A phased plan (metrics + advisory first, actuation later) would make review easier. <!-- streview-comment:56 --> -- 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]
