feat(MCO): update charts, samples for MCO v0.1.0 (#58)

apeabody · web-flow · commit 28f9df647a95 · 2025-05-15T13:23:37.000-07:00
diff --git a/fleet-charts/argocd-clusterprofile-syncer/Chart.yaml b/fleet-charts/argocd-clusterprofile-syncer/Chart.yaml
@@ -14,7 +14,7 @@
 
 apiVersion: v2
 name: argocd-clusterprofile-syncer
-version: 0.0.1
+version: 0.1.0
 description: Argo CD ClusterProfile Syncer
 type: application
-appVersion: "0.0.1"
+appVersion: "0.1.0"
diff --git a/fleet-charts/argocd-clusterprofile-syncer/values.yaml b/fleet-charts/argocd-clusterprofile-syncer/values.yaml
@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-image: "us-docker.pkg.dev/gke-fleet-management/multicluster-orchestrator/argocd-syncer:v0.0.1"
+image: "us-docker.pkg.dev/gke-fleet-management/multicluster-orchestrator/argocd-syncer:v0.1.0"
diff --git a/fleet-charts/argocd-mco-plugin/Chart.yaml b/fleet-charts/argocd-mco-plugin/Chart.yaml
@@ -14,7 +14,7 @@
 
 apiVersion: v2
 name: argocd-mco-plugin
-version: 0.0.1
+version: 0.1.0
 description: Argo CD MCO Plugin
 type: application
-appVersion: "0.0.1"
+appVersion: "0.1.0"
diff --git a/fleet-charts/argocd-mco-plugin/values.yaml b/fleet-charts/argocd-mco-plugin/values.yaml
@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-image: "us-docker.pkg.dev/gke-fleet-management/multicluster-orchestrator/argocd-placement-plugin:v0.0.1"
+image: "us-docker.pkg.dev/gke-fleet-management/multicluster-orchestrator/argocd-placement-plugin:v0.1.0"
diff --git a/fleet-charts/orchestrator/Chart.yaml b/fleet-charts/orchestrator/Chart.yaml
@@ -14,7 +14,7 @@
 
 apiVersion: v2
 name: orchestrator
-version: 0.0.4
+version: 0.1.0
 description: Multi Cluster Orchestrator
 type: application
-appVersion: "0.0.1"
+appVersion: "0.1.0"
diff --git a/fleet-charts/orchestrator/templates/crd/orchestra.multicluster.x-k8s.io_multikubernetesclusterplacements.yaml b/fleet-charts/orchestrator/templates/crd/orchestra.multicluster.x-k8s.io_multikubernetesclusterplacements.yaml
@@ -4,6 +4,7 @@ kind: CustomResourceDefinition
 metadata:
   annotations:
     controller-gen.kubebuilder.io/version: v0.16.1
+    helm.sh/resource-policy: keep
   name: multikubernetesclusterplacements.orchestra.multicluster.x-k8s.io
 spec:
   group: orchestra.multicluster.x-k8s.io
@@ -41,14 +42,14 @@ spec:
             description: MultiKubernetesClusterPlacementSpec defines the desired state
               of MultiKubernetesClusterPlacement
             properties:
-              rules:
+              clusterSelectorRules:
                 description: |-
-                  Rules defines a pipeline of rules which consume an ordered list of
-                  clusters and output an ordered list of clusters to generate the list of
+                  ClusterSelectorRules defines a pipeline of rules which consume an ordered list
+                  of clusters and output an ordered list of clusters to generate the list of
                   current target clusters.
                 items:
                   description: |-
-                    MultiKubernetesClusterPlacementRule defines a rule which takes an ordered list of
+                    PlacementClusterSelectorRule defines a rule which takes an ordered list of
                     clusters and returns a list of clusters based on the rule type and arguments.
                   properties:
                     arguments:
@@ -66,6 +67,7 @@ spec:
                         Type specifies the rule type and may be one of:
                         - all-clusters: all clusters defined by the ClusterProfiles present
                         - cluster-list: a user-provided ordered list of clusters
+                        - cluster-name-regex: a user-provided regular expression to match cluster names against
                       enum:
                       - all-clusters
                       - cluster-list
@@ -127,7 +129,7 @@ spec:
                     type: object
                 type: object
             required:
-            - rules
+            - clusterSelectorRules
             type: object
           status:
             description: MultiKubernetesClusterPlacementStatus defines the observed
@@ -141,6 +143,31 @@ spec:
                     PlacementCluster describes a cluster on which the workload should be placed.
                     This is consumed by workload delilvery systems.
                   properties:
+                    atCapacityCeiling:
+                      description: 'Whether the cluster is currently at its capacity
+                        ceiling. For informational/debugging purposes only. TODO:
+                        remove this'
+                      type: boolean
+                    draining:
+                      properties:
+                        currentReplicaCount:
+                          description: 'For informational/debugging purposes only.
+                            TODO: remove this'
+                          type: integer
+                        desiredMaxReplicas:
+                          description: Desired maximum replicas for a draining cluster.
+                            The drainer plugin should reconcile the HPA's max replicas
+                            field to this value.
+                          type: integer
+                        lastReplicaCountDecrease:
+                          description: The last time DesiredMaxReplicas was decreased.
+                          format: date-time
+                          type: string
+                      required:
+                      - currentReplicaCount
+                      - desiredMaxReplicas
+                      - lastReplicaCountDecrease
+                      type: object
                     lastTransitionTime:
                       format: date-time
                       type: string
diff --git a/fleet-charts/orchestrator/values.yaml b/fleet-charts/orchestrator/values.yaml
@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-image: "us-docker.pkg.dev/gke-fleet-management/multicluster-orchestrator/controller:v0.0.1"
+image: "us-docker.pkg.dev/gke-fleet-management/multicluster-orchestrator/controller:v0.1.0"
diff --git a/multi-cluster-orchestrator/samples/gemma_vllm/1-infrastructure/main.tf b/multi-cluster-orchestrator/samples/gemma_vllm/1-infrastructure/main.tf
@@ -28,6 +28,7 @@ data "google_project" "default" {}
 ### Enable Services
 resource "google_project_service" "default" {
   for_each = toset([
+    "cloudresourcemanager.googleapis.com",
     "compute.googleapis.com",
     "container.googleapis.com",
     "gkehub.googleapis.com",
@@ -63,6 +64,8 @@ resource "google_project_iam_member" "clusters" {
         for role in [
           "roles/container.defaultNodeServiceAccount",
           "roles/monitoring.metricWriter",
+          # This can be removed if nodes don't need to pull images from local Artifact Registry
+          "roles/artifactregistry.reader",
           # For image streaming
           "roles/serviceusage.serviceUsageConsumer"
         ] :
@@ -163,6 +166,8 @@ resource "google_container_cluster" "hub" {
   # Set `deletion_protection` to `true` will ensure that one cannot
   # accidentally delete this instance by use of Terraform.
   deletion_protection = false
+
+  depends_on = [google_project_service.default, google_project_iam_member.clusters["hub"]]
 }
 
 # Apply label to membership without importing the membership
@@ -221,6 +226,8 @@ resource "google_container_cluster" "clusters" {
   # Set `deletion_protection` to `true` will ensure that one cannot
   # accidentally delete this instance by use of Terraform.
   deletion_protection = false
+
+  depends_on = [google_project_service.default, google_project_iam_member.clusters["worker"]]
 }
 
 resource "google_container_node_pool" "gpu-node-pool" {
@@ -354,7 +361,7 @@ resource "helm_release" "orchestrator" {
   name       = "orchestrator"
   repository = "https://googlecloudplatform.github.io/gke-fleet-management"
   chart      = "orchestrator"
-  version    = "0.0.4"
+  version    = "0.1.0"
 
   lint = true
 
@@ -365,7 +372,7 @@ resource "helm_release" "argocd-clusterprofile-syncer" {
   name       = "argocd-clusterprofile-syncer"
   repository = "https://googlecloudplatform.github.io/gke-fleet-management"
   chart      = "argocd-clusterprofile-syncer"
-  version    = "0.0.1"
+  version    = "0.1.0"
 
   # Deploy into the same namespace as ArgoCD
   namespace = helm_release.argocd.namespace
@@ -379,7 +386,7 @@ resource "helm_release" "argocd-mco-plugin" {
   name       = "argocd-mco-plugin"
   repository = "https://googlecloudplatform.github.io/gke-fleet-management"
   chart      = "argocd-mco-plugin"
-  version    = "0.0.1"
+  version    = "0.1.0"
 
   # Deploy into the same namespace as ArgoCD
   namespace = helm_release.argocd.namespace
diff --git a/multi-cluster-orchestrator/samples/gemma_vllm/2-workload/charts/gemma-vllm-application/Chart.yaml b/multi-cluster-orchestrator/samples/gemma_vllm/2-workload/charts/gemma-vllm-application/Chart.yaml
@@ -14,6 +14,6 @@
 
 apiVersion: v2
 name: gemma-vllm-application
-version: 0.0.7
+version: 0.0.8
 description: Gemma 3 vLLM Application
 type: application
diff --git a/multi-cluster-orchestrator/samples/gemma_vllm/2-workload/charts/gemma-vllm-application/templates/applicationset.yaml b/multi-cluster-orchestrator/samples/gemma_vllm/2-workload/charts/gemma-vllm-application/templates/applicationset.yaml
@@ -54,3 +54,9 @@ spec:
           parameters:
             - name: hf_api_token
               value: "{{.Values.hf_api_token}}"
+            {{`{{- if ne .desiredMaxReplicas nil }}`}}
+            - name: minReplicas
+              value: "1"
+            - name: maxReplicas
+              value: "{{`{{ .desiredMaxReplicas }}`}}"
+            {{`{{- end }}`}}
diff --git a/multi-cluster-orchestrator/samples/gemma_vllm/2-workload/charts/gemma-vllm-application/templates/placement.yaml b/multi-cluster-orchestrator/samples/gemma_vllm/2-workload/charts/gemma-vllm-application/templates/placement.yaml
@@ -17,13 +17,14 @@ kind: MultiKubernetesClusterPlacement
 metadata:
   name: gemma-vllm-placement-autoscale
 spec:
-  rules:
+  clusterSelectorRules:
     - type: "all-clusters"
     - type: "cluster-name-regex"
       arguments:
         regex: "^fleet-cluster-inventory/{{ .Values.clusters_prefix }}-"
   scaling:
     autoscaleForCapacity:
+      useDraining: true
       minClustersBelowCapacityCeiling: 1
       workloadDetails:
         namespace: {{ .Release.Namespace }}
diff --git a/multi-cluster-orchestrator/samples/gemma_vllm/README.md b/multi-cluster-orchestrator/samples/gemma_vllm/README.md
@@ -38,6 +38,7 @@ Kubernetes clusters and streamline your DevOps workflows. To learn more, see
 1. [Make sure that billing is enabled for your Google Cloud project][3].
 
 1. The following service APIs will be enabled:
+   1. Cloud Resource Manager
    1. Compute Engine
    1. Google Kubernetes Engine
    1. GKE Hub
@@ -69,7 +70,7 @@ In this tutorial, you should use [Cloud Shell][7] to manage resources
 hosted on Google Cloud. Cloud Shell is preinstalled with the software you need
 for this tutorial, including [Terraform][8], [kubectl][9], and [gcloud CLI][10].
 
-> [!NOTE]
+> [!IMPORTANT]
 > If you do not use Cloud Shell, you may need to install
 > Terraform, kubectl, and gcloud CLI. You must also set your default
 > Terraform project with: `export GOOGLE_CLOUD_PROJECT=PROJECT_ID`.
diff --git a/multi-cluster-orchestrator/samples/gemma_vllm/images/workloads.png b/multi-cluster-orchestrator/samples/gemma_vllm/images/workloads.png
diff --git a/multi-cluster-orchestrator/samples/hello_world/1-infrastructure/main.tf b/multi-cluster-orchestrator/samples/hello_world/1-infrastructure/main.tf
@@ -27,6 +27,7 @@ data "google_project" "default" {}
 ### Enable Services
 resource "google_project_service" "default" {
   for_each = toset([
+    "cloudresourcemanager.googleapis.com",
     "compute.googleapis.com",
     "container.googleapis.com",
     "gkehub.googleapis.com",
@@ -62,6 +63,8 @@ resource "google_project_iam_member" "clusters" {
         for role in [
           "roles/container.defaultNodeServiceAccount",
           "roles/monitoring.metricWriter",
+          # This can be removed if nodes don't need to pull images from local Artifact Registry
+          "roles/artifactregistry.reader",
           # For image streaming
           "roles/serviceusage.serviceUsageConsumer"
         ] :
@@ -143,7 +146,7 @@ resource "google_container_cluster" "hub" {
   # accidentally delete this instance by use of Terraform.
   deletion_protection = false
 
-  depends_on = [google_project_service.default]
+  depends_on = [google_project_service.default, google_project_iam_member.clusters["hub"]]
 }
 
 # Apply label to membership without importing the membership
@@ -179,15 +182,15 @@ resource "google_container_cluster" "clusters" {
 
   cluster_autoscaling {
     auto_provisioning_defaults {
-      service_account = google_service_account.clusters["hub"].email
+      service_account = google_service_account.clusters["worker"].email
     }
   }
 
   # Set `deletion_protection` to `true` will ensure that one cannot
   # accidentally delete this instance by use of Terraform.
   deletion_protection = false
 
-  depends_on = [google_project_service.default]
+  depends_on = [google_project_service.default, google_project_iam_member.clusters["worker"]]
 }
 
 ### Multi-Cluster Gateway
@@ -286,7 +289,7 @@ resource "helm_release" "orchestrator" {
   name       = "orchestrator"
   repository = "https://googlecloudplatform.github.io/gke-fleet-management"
   chart      = "orchestrator"
-  version    = "0.0.4"
+  version    = "0.1.0"
 
   lint = true
 
@@ -297,7 +300,7 @@ resource "helm_release" "argocd-clusterprofile-syncer" {
   name       = "argocd-clusterprofile-syncer"
   repository = "https://googlecloudplatform.github.io/gke-fleet-management"
   chart      = "argocd-clusterprofile-syncer"
-  version    = "0.0.1"
+  version    = "0.1.0"
 
   # Deploy into the same namespace as ArgoCD
   namespace = helm_release.argocd.namespace
@@ -311,7 +314,7 @@ resource "helm_release" "argocd-mco-plugin" {
   name       = "argocd-mco-plugin"
   repository = "https://googlecloudplatform.github.io/gke-fleet-management"
   chart      = "argocd-mco-plugin"
-  version    = "0.0.1"
+  version    = "0.1.0"
 
   # Deploy into the same namespace as ArgoCD
   namespace = helm_release.argocd.namespace
diff --git a/multi-cluster-orchestrator/samples/hello_world/2-workload/charts/hello-world-application/Chart.yaml b/multi-cluster-orchestrator/samples/hello_world/2-workload/charts/hello-world-application/Chart.yaml
@@ -14,6 +14,6 @@
 
 apiVersion: v2
 name: hello-world-application
-version: 0.0.1
+version: 0.0.2
 description: Hello World Application
 type: application
diff --git a/multi-cluster-orchestrator/samples/hello_world/2-workload/charts/hello-world-application/templates/applicationset.yaml b/multi-cluster-orchestrator/samples/hello_world/2-workload/charts/hello-world-application/templates/applicationset.yaml
@@ -18,31 +18,42 @@ metadata:
   name: hello-world-applicationset
   namespace: argocd
 spec:
-    generators:
-    - plugin:
-        configMapRef:
-          name: argocd-mco-placement
-        input:
+  generators:
+  - plugin:
+      configMapRef:
+        name: argocd-mco-placement
+      input:
+        parameters:
+          multiClusterOrchestraPlacementName: "hello-web-placement-autoscale"
+          multiClusterOrchestraPlacementNamespace: {{ .Release.Namespace }}
+      requeueAfterSeconds: 10
+  syncPolicy:
+    applicationsSync: sync
+    preserveResourcesOnDeletion: false
+  template:
+    metadata:
+      name: '{{`{{name}}`}}-hw'
+    spec:
+      destination:
+        namespace: {{ .Release.Namespace }}
+        name: '{{`{{name}}`}}'
+      syncPolicy:
+        automated: {}
+        syncOptions:
+          - CreateNamespace=true
+      project: default
+      source:
+        path: fleet-charts/hello-world
+        repoURL: https://github.com/GoogleCloudPlatform/gke-fleet-management.git
+        targetRevision: HEAD
+  templatePatch: |
+    spec:
+      source:
+        helm:
           parameters:
-            multiClusterOrchestraPlacementName: "hello-web-placement-autoscale"
-            multiClusterOrchestraPlacementNamespace: {{ .Release.Namespace }}
-        requeueAfterSeconds: 10
-    syncPolicy:
-      applicationsSync: sync
-      preserveResourcesOnDeletion: false
-    template:
-      metadata:
-        name: '{{`{{name}}`}}-hw'
-      spec:
-        destination:
-          namespace: {{ .Release.Namespace }}
-          name: '{{`{{name}}`}}'
-        syncPolicy:
-          automated: {}
-          syncOptions:
-            - CreateNamespace=true
-        project: default
-        source:
-          path: fleet-charts/hello-world
-          repoURL: https://github.com/GoogleCloudPlatform/gke-fleet-management.git
-          targetRevision: HEAD
+            {{`{{- if ne .desiredMaxReplicas nil }}`}}
+            - name: hpa.minReplicas
+              value: "1"
+            - name: hpa.maxReplicas
+              value: "{{`{{ .desiredMaxReplicas }}`}}"
+            {{`{{- end }}`}}
diff --git a/multi-cluster-orchestrator/samples/hello_world/2-workload/charts/hello-world-application/templates/placement.yaml b/multi-cluster-orchestrator/samples/hello_world/2-workload/charts/hello-world-application/templates/placement.yaml
@@ -24,6 +24,7 @@ spec:
         regex: "^fleet-cluster-inventory/{{ .Values.clusters_prefix }}-"
   scaling:
     autoscaleForCapacity:
+      useDraining: true
       minClustersBelowCapacityCeiling: 1
       workloadDetails:
         namespace: {{ .Release.Namespace }}
diff --git a/multi-cluster-orchestrator/samples/hello_world/README.md b/multi-cluster-orchestrator/samples/hello_world/README.md
