[Feature] Support recreate pods for RayCluster using RayClusterSpec

Signed-off-by: win5923 <[email protected]>

Author: win5923

docs/reference/api.md

@@ -277,6 +277,36 @@ _Appears in:_
 | `headGroupSpec` _[HeadGroupSpec](#headgroupspec)_ | HeadGroupSpec is the spec for the head pod |  |  |
 | `rayVersion` _string_ | RayVersion is used to determine the command for the Kubernetes Job managed by RayJob |  |  |
 | `workerGroupSpecs` _[WorkerGroupSpec](#workergroupspec) array_ | WorkerGroupSpecs are the specs for the worker pods |  |  |
+| `upgradeStrategy` _[RayClusterUpgradeStrategy](#rayclusterupgradestrategy)_ | UpgradeStrategy defines the strategy used when upgrading the RayCluster Pods. |  |  |
+
+
+#### RayClusterUpgradeStrategy
+
+
+
+
+
+
+
+_Appears in:_
+- [RayClusterSpec](#rayclusterspec)
+
+| Field | Description | Default | Validation |
+| --- | --- | --- | --- |
+| `type` _[RayClusterUpgradeType](#rayclusterupgradetype)_ | Type represents the strategy used when upgrading the RayCluster Pods. Currently supports `Recreate` and `None`. |  |  |
+
+
+#### RayClusterUpgradeType
+
+_Underlying type:_ _string_
+
+
+
+
+
+_Appears in:_
+- [RayClusterUpgradeStrategy](#rayclusterupgradestrategy)
+
 
 
 #### RayJob

helm-chart/kuberay-operator/crds/ray.io_rayclusters.yaml

@@ -9,6 +9,21 @@ import (
 // EDIT THIS FILE!  THIS IS SCAFFOLDING FOR YOU TO OWN!
 // NOTE: json tags are required.  Any new fields you add must have json tags for the fields to be serialized.
 
+type RayClusterUpgradeType string
+
+const (
+	// Recreate strategy allows deleting old Pods and automatically recreate them with the new spec
+	Recreate RayClusterUpgradeType = "Recreate"
+	// No new pod will be created while the strategy is set to None
+	RayClusterUpgradeNone RayClusterUpgradeType = "None"
+)
+
+type RayClusterUpgradeStrategy struct {
+	// Type represents the strategy used when upgrading the RayCluster Pods. Currently supports `Recreate` and `None`.
+	// +optional
+	Type *RayClusterUpgradeType `json:"type,omitempty"`
+}
+
 // RayClusterSpec defines the desired state of RayCluster
 type RayClusterSpec struct {
 	// Suspend indicates whether a RayCluster should be suspended.
@@ -44,6 +59,9 @@ type RayClusterSpec struct {
 	// WorkerGroupSpecs are the specs for the worker pods
 	// +optional
 	WorkerGroupSpecs []WorkerGroupSpec `json:"workerGroupSpecs,omitempty"`
+	// UpgradeStrategy defines the strategy used when upgrading the RayCluster Pods.
+	// +optional
+	UpgradeStrategy *RayClusterUpgradeStrategy `json:"upgradeStrategy,omitempty"`
 }
 
 // GcsFaultToleranceOptions contains configs for GCS FT

helm-chart/kuberay-operator/crds/ray.io_rayjobs.yaml

@@ -28,7 +28,7 @@ const (
 	// During upgrade, NewCluster strategy will create new upgraded cluster and switch to it when it becomes ready
 	NewCluster RayServiceUpgradeType = "NewCluster"
 	// No new cluster will be created while the strategy is set to None
-	None RayServiceUpgradeType = "None"
+	RayServiceUpgradeNone RayServiceUpgradeType = "None"
 )
 
 // These statuses should match Ray Serve's application statuses