DanielLeens commented on issue #12087: URL: https://github.com/apache/seatunnel/issues/12087#issuecomment-5551704457
Thanks @CryoThrust. I rechecked current dev and the open PR set; there is still no AutoScaler implementation or overlapping autoscaling PR. The first slice can proceed as contract and policy work only. Please place it in `seatunnel-engine/seatunnel-engine-server` under a new server-local `org.apache.seatunnel.engine.server.autoscale` package. `CoordinatorService` already owns active-master scheduling state, while `ResourceManager` exposes registered worker, slot, and resource information; the first PR should not introduce a public engine API or REST and Prometheus adapter yet. Keep the immutable snapshot and recommendation models, pure policy evaluator, and monotonic-clock stabilization tracker together in that package. The required deterministic tests are epoch and generation fencing, duplicate advisory idempotency, Fixed versus Dynamic Slot semantics, stale and future worker samples, missing metrics blocking scale-in, signal conflicts, and active-master handover reset. A later integration PR can project that tested model into REST and metrics without duplicating decision logic. Please keep worker mutation, drain, actuator calls, and running-job rescale out of this first slice. -- 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]
