feat from kubernetes/autoscaler/cluster-autoscaler-release-1.28

Jenkinsfile

@@ -0,0 +1,88 @@
+// throttle concurrent build
+properties([[$class: 'ThrottleJobProperty', categories: [], limitOneJobWithMatchingParams: false, maxConcurrentPerNode: 1, maxConcurrentTotal: 1, paramsToUseForLimit: '', throttleEnabled: true, throttleOption: 'project']])
+
+library(identifier: "visenze-lib@${params['VISENZE_LIB_BRANCH'] ?: 'master'}", changelog: false)
+
+pipeline {
+  agent {
+    label "build-amd64"
+  }
+
+  options {
+    ansiColor('xterm')
+  }
+
+  tools {
+    go 'go1.20'
+  }
+
+  stages {
+    stage('Checkout') {
+      steps {
+        checkout([
+            $class: 'GitSCM',
+            branches: scm.branches,
+            doGenerateSubmoduleConfigurations: scm.doGenerateSubmoduleConfigurations,
+            extensions: [
+                [
+                    $class: 'CloneOption',
+                    noTags: true,
+                    reference: '',
+                    timeout: 60
+                ],
+                [
+                    $class: 'SubmoduleOption',
+                    disableSubmodules: false,
+                    parentCredentials: true,
+                    recursiveSubmodules: true,
+                    trackingSubmodules: true,
+                    reference: '',
+                    timeout: 60
+                ],
+                [$class: 'CleanBeforeCheckout']
+            ],
+            userRemoteConfigs: scm.userRemoteConfigs
+        ])
+      }
+    }
+
+    stage('Test') {
+      when {
+        expression {
+          return canRun()
+        }
+      }
+      steps {
+        script {
+          dir('cluster-autoscaler') {
+            sh('make test-in-docker')
+          }
+        }
+      }
+    }
+
+    stage('Docker Build&Push') {
+      when {
+        expression {
+          return canRun()
+        }
+      }
+      steps {
+        script {
+          dir('cluster-autoscaler') {
+            def version = sh(script: "grep ClusterAutoscalerVersion version/version.go",
+                returnStdout: true).split('"')[-2]
+            docker.withRegistry('', 'docker-hub-credential') {
+              sh("docker buildx create --use")
+              sh("docker buildx build -t visenze/cluster-autoscaler:${version} --push --platform linux/arm64,linux/amd64 .")
+            }
+          }
+        }
+      }
+    }
+  }
+}
+
+def canRun() {
+  return env.BRANCH_NAME.startsWith('release-') || env.BRANCH_NAME == 'master'
+}

cluster-autoscaler/Dockerfile

@@ -0,0 +1,21 @@
+FROM golang:1.20.14 AS build
+
+ARG TARGETARCH
+ENV GOPATH /gopath/
+ENV PATH $GOPATH/bin:$PATH
+ENV GO111MODULE auto
+ENV GOARCH ${TARGETARCH}
+
+RUN apt-get update && apt-get --yes install libseccomp-dev
+RUN go version
+RUN go get github.com/tools/godep
+RUN godep version
+
+WORKDIR /gopath/src/k8s.io/autoscaler/cluster-autoscaler
+ADD . .
+RUN CGO_ENABLED=0 GOOS=linux go build -o cluster-autoscaler --ldflags "-s"
+
+FROM alpine
+COPY --from=build /gopath/src/k8s.io/autoscaler/cluster-autoscaler/cluster-autoscaler /
+
+CMD ["./cluster-autoscaler"]

cluster-autoscaler/cloudprovider/aws/aws_manager.go

@@ -39,6 +39,8 @@ import (
 	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go/service/eks"
 	"k8s.io/autoscaler/cluster-autoscaler/config"
 	"k8s.io/autoscaler/cluster-autoscaler/utils/gpu"
+	"k8s.io/autoscaler/cluster-autoscaler/utils/gpumemory"
+	"k8s.io/autoscaler/cluster-autoscaler/utils/mpscontext"
 )
 
 const (
@@ -279,6 +281,8 @@ func (m *AwsManager) buildNodeFromTemplate(asg *asg, template *asgTemplate) (*ap
 	node.Status.Capacity[apiv1.ResourceCPU] = *resource.NewQuantity(template.InstanceType.VCPU, resource.DecimalSI)
 	node.Status.Capacity[gpu.ResourceNvidiaGPU] = *resource.NewQuantity(template.InstanceType.GPU, resource.DecimalSI)
 	node.Status.Capacity[apiv1.ResourceMemory] = *resource.NewQuantity(template.InstanceType.MemoryMb*1024*1024, resource.DecimalSI)
+	node.Status.Capacity[gpumemory.ResourceVisenzeGPUMemory] = *resource.NewQuantity(template.InstanceType.GPUMemory, resource.DecimalSI)
+	node.Status.Capacity[mpscontext.ResourceVisenzeMPSContext] = *resource.NewQuantity(template.InstanceType.MPSContext, resource.DecimalSI)
 
 	m.updateCapacityWithRequirementsOverrides(&node.Status.Capacity, asg.MixedInstancesPolicy)
 

cluster-autoscaler/cloudprovider/aws/ec2_instance_types.go

@@ -25,6 +25,8 @@ type InstanceType struct {
 	MemoryMb     int64
 	GPU          int64
 	Architecture string
+	GPUMemory    int64
+	MPSContext   int64
 }
 
 // StaticListLastUpdateTime is a string declaring the last time the static list was updated.

cluster-autoscaler/cloudprovider/builder/builder_all.go

@@ -49,6 +49,7 @@ import (
 	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/tencentcloud"
 	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/volcengine"
 	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/vultr"
+	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/spotinst"
 	"k8s.io/autoscaler/cluster-autoscaler/config"
 )
 
@@ -143,6 +144,8 @@ func buildCloudProvider(opts config.AutoscalingOptions, do cloudprovider.NodeGro
 		return scaleway.BuildScaleway(opts, do, rl)
 	case cloudprovider.RancherProviderName:
 		return rancher.BuildRancher(opts, do, rl)
+	case cloudprovider.SpotinstProviderName:
+		return spotinst.BuildSpotinst(opts, do, rl)
 	case cloudprovider.VolcengineProviderName:
 		return volcengine.BuildVolcengine(opts, do, rl)
 	}

cluster-autoscaler/cloudprovider/cloud_provider.go

@@ -86,6 +86,8 @@ const (
 	CivoProviderName = "civo"
 	// RancherProviderName gets the provider name of rancher
 	RancherProviderName = "rancher"
+	// SpotinstProviderName gets the provider name of aws
+	SpotinstProviderName = "spotinst"
 )
 
 // GpuConfig contains the label, type and the resource name for a GPU.