Add custom interpreter for Raycluster

pkg/resourceinterpreter/default/thirdparty/resourcecustomizations/ray.io/v1/RayCluster/customizations.yaml

@@ -0,0 +1,170 @@
+apiVersion: config.karmada.io/v1alpha1
+kind: ResourceInterpreterCustomization
+metadata:
+  name: declarative-configuration-raycluster
+spec:
+  target:
+    apiVersion: ray.io/v1
+    kind: RayCluster
+  customizations:
+    componentResource:
+      luaScript: |
+        local kube = require("kube")
+
+        local function isempty(s)
+          return s == nil or s == ''
+        end
+
+        function GetComponents(desiredObj)
+          local components = {}
+
+          -- Head component (always 1 replica)
+          if desiredObj.spec ~= nil and desiredObj.spec.headGroupSpec ~= nil and desiredObj.spec.headGroupSpec.template ~= nil then
+            local headRequires = kube.accuratePodRequirements(desiredObj.spec.headGroupSpec.template)
+            local headComponent = {
+              name = "ray-head",
+              replicas = 1,
+              replicaRequirements = headRequires
+            }
+            table.insert(components, headComponent)
+          end
+
+          -- Worker group components
+          if desiredObj.spec ~= nil and desiredObj.spec.workerGroupSpecs ~= nil then
+            for i, workerGroup in ipairs(desiredObj.spec.workerGroupSpecs) do
+              local name = workerGroup.groupName
+              if isempty(name) then
+                name = "worker-" .. tostring(i)
+              end
+              local replicas = workerGroup.replicas or 0
+              local requires = nil
+              if workerGroup.template ~= nil then
+                requires = kube.accuratePodRequirements(workerGroup.template)
+              end
+              local wgComponent = {
+                name = name,
+                replicas = replicas,
+                replicaRequirements = requires
+              }
+              table.insert(components, wgComponent)
+            end
+          end
+
+          return components
+        end
+    replicaResource:
+      luaScript: >
+        local kube = require("kube")
+
+        local function isempty(s)
+          return s == nil or s == ''
+        end
+
+        function GetReplicas(desiredObj)
+          -- Head pod is always 1 for RayCluster
+          local totalReplicas = 1
+          local requires = {
+            resourceRequest = {},
+            nodeClaim = {},
+          }
+
+          -- Initialize with head group resources
+          -- Until multiple podTemplates are supported, interpreter will only take affinity, toleration, and priorityclass input from head group to common podTemplate
+          requires.resourceRequest.cpu = desiredObj.spec.headGroupSpec.template.spec.containers[1].resources.limits.cpu
+          requires.resourceRequest.memory = desiredObj.spec.headGroupSpec.template.spec.containers[1].resources.limits.memory
+          if not isempty(desiredObj.spec.headGroupSpec.template.spec.nodeSelector) then
+            requires.nodeClaim.nodeSelector = desiredObj.spec.headGroupSpec.template.spec.nodeSelector
+          end
+          if not isempty(desiredObj.spec.headGroupSpec.template.spec.tolerations) then
+            requires.nodeClaim.tolerations = desiredObj.spec.headGroupSpec.template.spec.tolerations
+          end
+          if not isempty(desiredObj.spec.headGroupSpec.template.spec.priorityClassName) then
+            requires.priorityClassName = desiredObj.spec.headGroupSpec.template.spec.priorityClassName
+          end
+
+          -- Convert to numeric for comparison
+          local max_memory_value = kube.getResourceQuantity(requires.resourceRequest.memory)
+
+          -- Add replicas from all worker group specs 
+          -- Until multiple podTemplates are supported in replicaRequirements, take max of cpu + memory values as requirement
+          if desiredObj.spec.workerGroupSpecs then
+            for i, workerGroup in ipairs(desiredObj.spec.workerGroupSpecs) do
+              if workerGroup.replicas then
+                totalReplicas = totalReplicas + workerGroup.replicas
+              end
+              if not isempty(workerGroup.template.spec.containers.resources.limits.cpu) then
+                requires.resourceRequest.cpu = math.max(requires.resourceRequest.cpu, workerGroup.template.spec.containers.resources.limits.cpu)
+              end
+              if not isempty(workerGroup.template.spec.containers.resources.limits.memory) then
+                local memory_value = kube.getResourceQuantity(workerGroup.template.spec.containers.resources.limits.memory)
+                if memory_value > max_memory_value then
+                  max_memory_value = memory_value
+                  requires.resourceRequest.memory = workerGroup.template.spec.containers.resources.limits.memory
+                end
+              end
+            end
+          end
+
+          return totalReplicas, requires
+        end
+    healthInterpretation:
+      luaScript: >
+        function InterpretHealth(observedObj)
+          if observedObj.status == nil or observedObj.status.state == nil then
+            return false
+          end
+          return observedObj.status.state == 'ready'
+        end
+    statusAggregation:
+      luaScript: >
+        function AggregateStatus(desiredObj, statusItems)
+          if desiredObj.status == nil then
+            desiredObj.status = {}
+          end
+
+          for i = 1, #statusItems do
+            currentStatus = statusItems[i].status
+            if currentStatus ~= nil then
+              desiredObj.status.state = currentStatus.state
+              desiredObj.status.stateTransitionTimes = currentStatus.stateTransitionTimes
+              desiredObj.status.conditions = currentStatus.conditions
+              desiredObj.status.desiredCPU = currentStatus.desiredCPU
+              desiredObj.status.desiredGPU = currentStatus.desiredGPU
+              desiredObj.status.desiredMemory = currentStatus.desiredMemory
+              desiredObj.status.desiredTPU = currentStatus.desiredTPU
+              desiredObj.status.endpoints = currentStatus.endpoints
+              desiredObj.status.head = currentStatus.head
+              desiredObj.status.lastUpdateTime = currentStatus.lastUpdateTime
+              desiredObj.status.maxWorkerReplicas = currentStatus.maxWorkerReplicas
+              desiredObj.status.minWorkerReplicas = currentStatus.minWorkerReplicas
+              desiredObj.status.observedGeneration = currentStatus.observedGeneration
+              desiredObj.status.readyWorkerReplicas = currentStatus.readyWorkerReplicas
+            end
+          end
+
+          return desiredObj
+        end
+    statusReflection:
+      luaScript: >
+        function ReflectStatus(observedObj)
+          status = {}
+          if observedObj == nil or observedObj.status == nil then
+            return status
+          end
+          status.state = observedObj.status.state
+          status.stateTransitionTimes = observedObj.status.stateTransitionTimes
+          status.conditions = observedObj.status.conditions
+          status.desiredCPU = observedObj.status.desiredCPU
+          status.desiredGPU = observedObj.status.desiredGPU
+          status.desiredMemory = observedObj.status.desiredMemory
+          status.desiredTPU = observedObj.status.desiredTPU
+          status.desiredWorkerReplicas = observedObj.status.desiredWorkerReplicas
+          status.endpoints = observedObj.status.endpoints
+          status.head = observedObj.status.head
+          status.lastUpdateTime = observedObj.status.lastUpdateTime
+          status.maxWorkerReplicas = observedObj.status.maxWorkerReplicas
+          status.minWorkerReplicas = observedObj.status.minWorkerReplicas
+          status.observedGeneration = observedObj.status.observedGeneration
+          status.readyWorkerReplicas = observedObj.status.readyWorkerReplicas
+          return status
+        end

pkg/resourceinterpreter/default/thirdparty/resourcecustomizations/ray.io/v1/RayCluster/customizations_tests.yaml

@@ -0,0 +1,12 @@
+tests:
+  - desiredInputPath: testdata/desired-raycluster.yaml
+    statusInputPath: testdata/status-file.yaml
+    operation: AggregateStatus
+  - desiredInputPath: testdata/desired-raycluster.yaml
+    operation: InterpretReplica
+  - observedInputPath: testdata/observed-raycluster.yaml
+    operation: InterpretHealth
+  - observedInputPath: testdata/observed-raycluster.yaml
+    operation: InterpretStatus
+  - observedInputPath: testdata/observed-raycluster.yaml
+    operation: InterpretComponent

pkg/resourceinterpreter/default/thirdparty/resourcecustomizations/ray.io/v1/RayCluster/testdata/desired-raycluster.yaml

@@ -0,0 +1,44 @@
+apiVersion: ray.io/v1
+kind: RayCluster
+metadata:
+  name: sample
+spec:
+  rayVersion: '2.46.0'
+  headGroupSpec:
+    rayStartParams: {}
+    template:
+      spec:
+        containers:
+        - name: ray-head
+          image: rayproject/ray:2.46.0
+          resources:
+            limits:
+              cpu: 1
+              memory: 2G
+            requests:
+              cpu: 1
+              memory: 2G
+          ports:
+          - containerPort: 6379
+            name: gcs-server
+          - containerPort: 8265
+            name: dashboard
+          - containerPort: 10001
+            name: client
+  workerGroupSpecs:
+  - replicas: 1
+    minReplicas: 1
+    maxReplicas: 5
+    groupName: workergroup
+    rayStartParams: {}
+    template:
+      spec:
+        containers:
+          image: rayproject/ray:2.46.0
+          resources:
+            limits:
+              cpu: 1
+              memory: 1G
+            requests:
+              cpu: 1
+              memory: 1G

pkg/resourceinterpreter/default/thirdparty/resourcecustomizations/ray.io/v1/RayCluster/testdata/observed-raycluster.yaml

@@ -0,0 +1,103 @@
+apiVersion: ray.io/v1
+kind: RayCluster
+metadata:
+  annotations:
+    kubectl.kubernetes.io/last-applied-configuration: |
+      {"apiVersion":"ray.io/v1","kind":"RayCluster","metadata":{"annotations":{},"name":"sample","namespace":"default"},"spec":{"headGroupSpec":{"rayStartParams":{},"template":{"spec":{"containers":[{"image":"rayproject/ray:2.46.0","name":"ray-head","ports":[{"containerPort":6379,"name":"gcs-server"},{"containerPort":8265,"name":"dashboard"},{"containerPort":10001,"name":"client"}],"resources":{"limits":{"cpu":1,"memory":"2G"},"requests":{"cpu":1,"memory":"2G"}}}]}}},"rayVersion":"2.46.0","workerGroupSpecs":[{"groupName":"workergroup","maxReplicas":5,"minReplicas":1,"rayStartParams":{},"replicas":1,"template":{"spec":{"containers":[{"image":"rayproject/ray:2.46.0","name":"ray-worker","resources":{"limits":{"cpu":1,"memory":"1G"},"requests":{"cpu":1,"memory":"1G"}}}]}}}]}}
+  creationTimestamp: "2025-09-21T03:54:44Z"
+  generation: 1
+  name: sample
+  namespace: default
+  resourceVersion: "850"
+  uid: 040acb09-4e53-4a23-a8b2-28b4300af70b
+spec:
+  headGroupSpec:
+    rayStartParams: {}
+    template:
+      spec:
+        containers:
+        - image: rayproject/ray:2.46.0
+          name: ray-head
+          ports:
+          - containerPort: 6379
+            name: gcs-server
+            protocol: TCP
+          - containerPort: 8265
+            name: dashboard
+            protocol: TCP
+          - containerPort: 10001
+            name: client
+            protocol: TCP
+          resources:
+            limits:
+              cpu: 1
+              memory: 2G
+            requests:
+              cpu: 1
+              memory: 2G
+  rayVersion: 2.46.0
+  workerGroupSpecs:
+  - groupName: workergroup
+    maxReplicas: 5
+    minReplicas: 1
+    numOfHosts: 1
+    rayStartParams: {}
+    replicas: 1
+    template:
+      spec:
+        containers:
+        - image: rayproject/ray:2.46.0
+          name: ray-worker
+          resources:
+            limits:
+              cpu: 1
+              memory: 1G
+            requests:
+              cpu: 1
+              memory: 1G
+status:
+  availableWorkerReplicas: 1
+  conditions:
+  - lastTransitionTime: "2025-09-21T03:55:30Z"
+    message: ""
+    reason: HeadPodRunningAndReady
+    status: "True"
+    type: HeadPodReady
+  - lastTransitionTime: "2025-09-21T03:55:45Z"
+    message: All Ray Pods are ready for the first time
+    reason: AllPodRunningAndReadyFirstTime
+    status: "True"
+    type: RayClusterProvisioned
+  - lastTransitionTime: "2025-09-21T03:54:44Z"
+    message: ""
+    reason: RayClusterSuspended
+    status: "False"
+    type: RayClusterSuspended
+  - lastTransitionTime: "2025-09-21T03:54:44Z"
+    message: ""
+    reason: RayClusterSuspending
+    status: "False"
+    type: RayClusterSuspending
+  desiredCPU: "2"
+  desiredGPU: "0"
+  desiredMemory: 3G
+  desiredTPU: "0"
+  desiredWorkerReplicas: 1
+  endpoints:
+    client: "10001"
+    dashboard: "8265"
+    gcs-server: "6379"
+    metrics: "8080"
+  head:
+    podIP: 10.244.0.6
+    podName: sample-head-9cvfc
+    serviceIP: 10.244.0.6
+    serviceName: sample-head-svc
+  lastUpdateTime: "2025-09-21T03:55:45Z"
+  maxWorkerReplicas: 5
+  minWorkerReplicas: 1
+  observedGeneration: 1
+  readyWorkerReplicas: 1
+  state: ready
+  stateTransitionTimes:
+    ready: "2025-09-21T03:55:45Z"

pkg/resourceinterpreter/default/thirdparty/resourcecustomizations/ray.io/v1/RayCluster/testdata/status-file.yaml

@@ -0,0 +1,48 @@
+applied: true
+clusterName: member1
+status:
+  availableWorkerReplicas: 1
+  conditions:
+  - lastTransitionTime: "2025-09-21T03:55:30Z"
+    message: ""
+    reason: HeadPodRunningAndReady
+    status: "True"
+    type: HeadPodReady
+  - lastTransitionTime: "2025-09-21T03:55:45Z"
+    message: All Ray Pods are ready for the first time
+    reason: AllPodRunningAndReadyFirstTime
+    status: "True"
+    type: RayClusterProvisioned
+  - lastTransitionTime: "2025-09-21T03:54:44Z"
+    message: ""
+    reason: RayClusterSuspended
+    status: "False"
+    type: RayClusterSuspended
+  - lastTransitionTime: "2025-09-21T03:54:44Z"
+    message: ""
+    reason: RayClusterSuspending
+    status: "False"
+    type: RayClusterSuspending
+  desiredCPU: "2"
+  desiredGPU: "0"
+  desiredMemory: 3G
+  desiredTPU: "0"
+  desiredWorkerReplicas: 1
+  endpoints:
+    client: "10001"
+    dashboard: "8265"
+    gcs-server: "6379"
+    metrics: "8080"
+  head:
+    podIP: 10.244.0.6
+    podName: sample-head-9cvfc
+    serviceIP: 10.244.0.6
+    serviceName: sample-head-svc
+  lastUpdateTime: "2025-09-21T03:55:45Z"
+  maxWorkerReplicas: 5
+  minWorkerReplicas: 1
+  observedGeneration: 1
+  readyWorkerReplicas: 1
+  state: ready
+  stateTransitionTimes:
+    ready: "2025-09-21T03:55:45Z"