[Umbrella] RayService Refactoring

[kevin85421]

Search before asking

• I had searched in the issues and found no similar feature requirement.

Description

The issue list is still WIP.

• [RayService][Refactor] RayService events #2547
• [RayService][Refactor] ServeConfigs memory leak #2549
• [RayService][Refactor] Change the data structure of ServeConfigs to avoid iterating the whole map  #2550
• [RayService][Refactor] Remove ctrlResult #2545
• [RayService][Refactor] Avoid flooding Kubernetes events #2546
• [RayService][refactor] Remove updateState #2705
• [RayService][refactor] Unify the logic of shouldPrepareNewRayCluster and createRayClusterInstanceIfNeeded #2706
• [RayService][refactor] RayService cluster watches Ingress #2707
• [RayService] Avoid sending health check requests to the head Pod when excludeHeadPodFromServeSvc is true #2776
• [RayService] Move the cluster switch logic from reconcileServe to Reconcile #2777
• [RayService] Add a safeguard and remove the dead code to ensure that both clusters are not empty before reconciling serve #2778
• [RayService] Always check the readiness of head Pods for both pending / active clusters if cluster exists #2783
• [RayService] Rename Restarting to PreparingNewCluster #2785
• [RayService] Unify the cluster switch over logic together #2786
• [RayService][Refactor] Add conditions to RayService #2751
• [RayService] Add an e2e test for the case that the active cluster's head Pod can't serve requests during the upgrade process #2787
• [RayService] Add logs and remove in-place update for the TestOldHeadPodFailDuringUpgrade e2e test #2819
• [Refactor][RayService] Add conditions to RayService #2807
• [RayService] make checkIfNeedSubmitServeApplications more unit testable #2822
• [RayService] Remove HealthLastUpdateTime from ServeDeploymentStatus #2825
• [RayService] Move the update of RayClusterStatus to calculateStatus #2826
• [RayService] Remove updateStatusForActiveCluster #2827
• [RayService] Add an e2e test that head Pod crashes without GCS FT #2829
• [RayService] Avoid passing RayServiceStatus to functions in reconcileServe #2828
• [RayService] Passing serve applications to calculateStatus and avoid calling Status().Update(...) inside reconcileServe #2831
• [RayService] refactor envtest by adding a util function rayServiceTemplate #2833
• [RayService] Move cleanUpRayClusterInstance from reconcileRayCluster to Reconcile #2838
• [RayService] Remove WaitForServeDeploymentReady #2842
• [RayService] e2e for redeploying RayServe application after recreating a new Head Pod #2834
• [RayService] Use Ready condition in e2e tests #2849
• [RayService] reword the comment on ServiceStatus = rayv1.Running #2848
• [RayService] Use Ready condition in e2e tests #2854
• [RayService] Remove the dependencies between constructRayClusterForRayService and the reconciler to make it more unit testable #2853
• [RayService] Refactor reconcileRayCluster to avoid updating CR status in the function #2859
• [RayService] Mark ServiceStatus as deprecated #2863
• [RayService] Add checks of RayService conditions in e2e tests #2864
• [RayService] Merge initConditions into calculateConditions #2866
• [RayService] Trim Redis Cleanup job less than 63 chars #2846
• [RayService] Add an envtest for RayService happy path #2868
• [RayService] Setting observedGeneration inside calculateStatus  #2869
• [RayService] Add unit tests for isZeroDowntimeUpgradeEnabled #2871
• [RayService] Add an envtest for autoscaler #2872
• [RayService] Create k8s events after creating/updating k8s resources #2873
• [RayService] Refactor createRayClusterInstance #2874
• [RayService] Refactor updateRayClusterInstance #2875
• [RayService] A new RayCluster should not be prepared if the serve service points to the existing pending cluster #2877

Use case

No response

Related issues

No response

Are you willing to submit a PR?

• Yes I am willing to submit a PR!