Cadc 11585 Added ray to the science-platform (keel-dev) for evaluation purpose.

deployment/k8s-config/ray/install-ray.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+helm -n cadc-ray install example-cluster ./ray
+

deployment/k8s-config/ray/ray/.helmignore

@@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/

deployment/k8s-config/ray/ray/Chart.yaml

@@ -0,0 +1,10 @@
+apiVersion: v2
+name: ray
+description: A Helm chart for deployments of Ray on Kubernetes.
+type: application
+
+# Chart version.
+version: 0.1.0
+
+# Ray version.
+appVersion: "latest"

deployment/k8s-config/ray/ray/templates/_helpers.tpl

@@ -0,0 +1,10 @@
+{{/*
+Compute clusterMaxWorkers as the sum of per-pod-type max workers.
+*/}}
+{{- define "ray.clusterMaxWorkers" -}}
+{{- $total := 0 }}
+{{- range .Values.podTypes }}
+{{- $total = add $total .maxWorkers }}
+{{- end }}
+{{- $total }}
+{{- end }}

deployment/k8s-config/ray/ray/templates/operator_cluster_scoped.yaml

@@ -0,0 +1,35 @@
+{{- if and (not .Values.namespacedOperator) (not .Values.clusterOnly) }}
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: ray-operator
+  namespace: {{ .Values.operatorNamespace }}
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      cluster.ray.io/component: operator
+  template:
+    metadata:
+      labels:
+        cluster.ray.io/component: operator
+    spec:
+      serviceAccountName: default
+      containers:
+      - name: ray
+        imagePullPolicy: Always
+        image: {{ .Values.operatorImage }}
+        command: ["ray-operator"]
+        env:
+        - name: AUTOSCALER_MAX_NUM_FAILURES
+          value: "inf"
+        resources:
+          requests:
+            cpu: 1
+            memory: 1Gi
+            ephemeral-storage: 1Gi
+          limits:
+            memory: 2Gi
+            cpu: 1
+{{- end }}

deployment/k8s-config/ray/ray/templates/operator_namespaced.yaml

@@ -0,0 +1,38 @@
+{{- if and (.Values.namespacedOperator) (not .Values.clusterOnly) }}
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: ray-operator
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      cluster.ray.io/component: operator
+  template:
+    metadata:
+      labels:
+        cluster.ray.io/component: operator
+    spec:
+      serviceAccountName: default
+      containers:
+      - name: ray
+        imagePullPolicy: Always
+        image: {{ .Values.operatorImage }}
+        command: ["ray-operator"]
+        env:
+        - name: RAY_OPERATOR_POD_NAMESPACE
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.namespace
+        - name: AUTOSCALER_MAX_NUM_FAILURES
+          value: "inf"
+        resources:
+          requests:
+            cpu: 1
+            memory: 1Gi
+            ephemeral-storage: 1Gi
+          limits:
+            memory: 2Gi
+            cpu: 1
+{{- end }}

deployment/k8s-config/ray/ray/templates/raycluster.yaml

@@ -0,0 +1,100 @@
+{{- if not .Values.operatorOnly }}
+apiVersion: cluster.ray.io/v1
+kind: RayCluster
+metadata:
+  name: {{ .Release.Name }}
+spec:
+  # The maximum number of workers nodes to launch in addition to the head node.
+  maxWorkers: {{ include "ray.clusterMaxWorkers" . }}
+  # The autoscaler will scale up the cluster faster with higher upscaling speed.
+  # E.g., if the task requires adding more nodes then autoscaler will gradually
+  # scale up the cluster in chunks of upscaling_speed*currently_running_nodes.
+  # This number should be > 0.
+  upscalingSpeed: {{ .Values.upscalingSpeed | default 1.0 }}
+  # If a node is idle for this many minutes, it will be removed.
+  idleTimeoutMinutes: {{ .Values.idleTimeoutMinutes | default 5 }}
+  # Specify the pod type for the ray head node (as configured below).
+  headPodType: {{ .Values.headPodType }}
+  # Specify the allowed pod types for this ray cluster and the resources they provide.
+  podTypes:
+    {{- range $key, $val := .Values.podTypes }}
+    - name: {{ $key }}
+      minWorkers: {{ $val.minWorkers | default 0}}
+      maxWorkers: {{ $val.maxWorkers | default 0}}
+      {{- if $val.rayResources }}
+      rayResources:
+        {{- toYaml $val.rayResources | nindent 8 }}
+      {{- end }}
+      podConfig:
+        apiVersion: v1
+        kind: Pod
+        metadata:
+          generateName: {{ kebabcase $key }}-
+        spec:
+          restartPolicy: Never
+          # This volume allocates shared memory for Ray to use for its plasma
+          # object store. If you do not provide this, Ray will fall back to
+          # /tmp which cause slowdowns if is not a shared memory volume.
+          volumes:
+          - name: dshm
+            emptyDir:
+              medium: Memory
+          containers:
+          - name: ray-node
+            imagePullPolicy: Always
+            image: {{ $.Values.image }}
+            # Do not change this command - it keeps the pod alive until it is
+            # explicitly killed.
+            command: ["/bin/bash", "-c", "--"]
+            args: ['trap : TERM INT; sleep infinity & wait;']
+            env:
+            - name: RAY_gcs_server_rpc_server_thread_num
+              value: "1"
+            ports:
+            - containerPort: 6379  # Redis port for Ray <= 1.10.0. GCS server port for Ray >= 1.11.0.
+            - containerPort: 10001  # Used by Ray Client
+            - containerPort: 8265  # Used by Ray Dashboard
+            - containerPort: 8000 # Used by Ray Serve
+
+            # This volume allocates shared memory for Ray to use for its plasma
+            # object store. If you do not provide this, Ray will fall back to
+            # /tmp which cause slowdowns if is not a shared memory volume.
+            volumeMounts:
+            - mountPath: /dev/shm
+              name: dshm
+            resources:
+              requests:
+                cpu: {{ .CPU }}
+                memory: {{ .memory }}
+              limits:
+                cpu: {{ .CPU }}
+                # The maximum memory that this pod is allowed to use. The
+                # limit will be detected by ray and split to use 10% for
+                # redis, 30% for the shared memory object store, and the
+                # rest for application memory. If this limit is not set and
+                # the object store size is not set manually, ray will
+                # allocate a very large object store in each pod that may
+                # cause problems for other pods.
+                memory: {{ .memory }}
+                {{- if .GPU }}
+                nvidia.com/gpu: {{ .GPU }}
+                {{- end }}
+          {{- if .nodeSelector }}
+          nodeSelector:
+          {{- toYaml $val.nodeSelector | nindent 12 }}
+          {{- end }}
+          {{- if $val.tolerations }}
+          tolerations:
+          {{- toYaml $val.tolerations | nindent 10 }}
+          {{- end }}
+    {{- end }}
+  # Commands to start Ray on the head node. You don't need to change this.
+  # Note dashboard-host is set to 0.0.0.0 so that Kubernetes can port forward.
+  headStartRayCommands:
+    - ray stop
+    - ulimit -n 65536; ray start --head --port=6379 --no-monitor --dashboard-host 0.0.0.0
+  # Commands to start Ray on worker nodes. You don't need to change this.
+  workerStartRayCommands:
+    - ray stop
+    - ulimit -n 65536; ray start --address=$RAY_HEAD_IP:6379
+{{- end }}

deployment/k8s-config/ray/ray/values.yaml

@@ -0,0 +1,111 @@
+# Default values for Ray.
+
+# RayCluster settings:
+
+# image is Ray image to use for the head and workers of this Ray cluster.
+# It's recommended to build custom dependencies for your workload into this image,
+# taking one of the offical `rayproject/ray` images as base.
+image: rayproject/ray:latest
+# The autoscaler will scale up the cluster faster with higher upscaling speed.
+# If the task requires adding more nodes then autoscaler will gradually
+# scale up the cluster in chunks of upscaling_speed*currently_running_nodes.
+# This number should be > 0.
+upscalingSpeed: 1.0
+# If a node is idle for this many minutes, it will be removed.
+idleTimeoutMinutes: 5
+# headPodType is the podType used for the Ray head node (as configured below).
+headPodType: rayHeadType
+# podTypes is the list of pod configurations available for use as Ray nodes.
+podTypes:
+    # The key for each podType is a user-defined string.
+    # Since we set headPodType: rayHeadType, the Ray head pod will use the configuration
+    # defined in this entry of podTypes:
+    rayHeadType:
+        # CPU is the number of CPUs used by this pod type.
+        # (Used for both requests and limits. Must be an integer, as Ray does not support fractional CPUs.)
+        CPU: 1
+        # memory is the memory used by this Pod type.
+        # (Used for both requests and limits.)
+        memory: 1Gi
+        # GPU is the number of NVIDIA GPUs used by this pod type.
+        # (Optional, requires GPU nodes with appropriate setup. See https://docs.ray.io/en/master/cluster/kubernetes-gpu.html)
+        GPU: 0
+        # rayResources is an optional string-int mapping signalling additional resources to Ray.
+        # "CPU", "GPU", and "memory" are filled automatically based on the above settings, but can be overriden;
+        # For example, rayResources: {"CPU": 0} can be used in the head podType to prevent Ray from scheduling tasks on the head.
+        # See https://docs.ray.io/en/master/advanced.html#dynamic-remote-parameters for an example of usage of custom resources in a Ray task.
+        rayResources: {}
+        # Optionally, set a node selector for this podType: https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#nodeselector
+        nodeSelector: {}
+
+        # tolerations for Ray pods of this podType (the head's podType in this case)
+        #   ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/
+        #   Note that it is often not necessary to manually specify tolerations for GPU
+        #   usage on managed platforms such as AKS, EKS, and GKE.
+        #   ref: https://docs.ray.io/en/master/cluster/kubernetes-gpu.html
+        tolerations: []
+        # - key: "nvidia.com/gpu"
+        #   operator: Exists
+        #   effect: NoSchedule
+
+    # The key for each podType is a user-defined string.
+    rayWorkerType:
+        # minWorkers is the minimum number of Ray workers of this pod type to keep running.
+        minWorkers: 2
+        # maxWorkers is the maximum number of Ray workers of this pod type to which Ray will scale.
+        maxWorkers: 3
+        # memory is the memory used by this Pod type.
+        # (Used for both requests and limits.)
+        memory: 1Gi
+        # CPU is the number of CPUs used by this pod type.
+        # (Used for both requests and limits. Must be an integer, as Ray does not support fractional CPUs.)
+        CPU: 1
+        # GPU is the number of NVIDIA GPUs used by this pod type.
+        # (Optional, requires GPU nodes with appropriate setup. See https://docs.ray.io/en/master/cluster/kubernetes-gpu.html)
+        GPU: 0
+        # rayResources is an optional string-int mapping signalling additional resources to Ray.
+        # "CPU", "GPU", and "memory" are filled automatically based on the above settings, but can be overriden;
+        # For example, rayResources: {"CPU": 0} can be used in the head podType to prevent Ray from scheduling tasks on the head.
+        # See https://docs.ray.io/en/master/advanced.html#dynamic-remote-parameters for an example of usage of custom resources in a Ray task.
+        rayResources: {}
+        # Optionally, set a node selector for this Pod type. See https://kubernetes.io/docs/concepts/scheduling-eviction/assign-pod-node/#nodeselector
+        nodeSelector: {}
+
+        # tolerations for Ray pods of this podType
+        #   ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/
+        #   Note that it is often not necessary to manually specify tolerations for GPU
+        #   usage on managed platforms such as AKS, EKS, and GKE.
+        #   ref: https://docs.ray.io/en/master/cluster/kubernetes-gpu.html
+        tolerations: []
+        # - key: nvidia.com/gpu
+        #   operator: Exists
+        #   effect: NoSchedule
+
+    # Optionally, define more worker podTypes
+    # rayWorkerType2:
+    #   minWorkers: 0
+    #   maxWorkers: 10
+    #   memory: ...
+
+# Operator settings:
+
+# operatorOnly - If true, will only set up the Operator with this release,
+# without launching a Ray cluster.
+operatorOnly: false
+# clusterOnly - If true, will only create a RayCluster resource with this release,
+# without setting up the Operator.
+# (Useful when launching multiple Ray clusters.)
+clusterOnly: false
+# namespacedOperator - If true, the operator is scoped to the Release namespace
+# and only manages RayClusters in that namespace.
+# By default, the operator is cluster-scoped and runs in the default namespace.
+namespacedOperator: true
+# operatorNamepsace - If using a cluster-scoped operator (namespacedOperator: false), set the namespace
+# in which to launch the operator.
+operatorNamespace: default
+# operatorImage - The image used in the operator deployment.
+# It is recommended to use one of the official `rayproject/ray` images for the operator.
+# It is recommended to use the same Ray version in the operator as in the Ray clusters managed
+# by the operator. In other words, the images specified under the fields `operatorImage` and `image`
+# should carry matching Ray versions.
+operatorImage: rayproject/ray:latest

deployment/k8s-config/ray/uninstall-ray.sh

@@ -0,0 +1,2 @@
+kubectl -n cadc-ray delete raycluster example-cluster
+helm -n cadc-ray uninstall example-cluster