fix: upgrade k8s 1.34, fix shm path, helm chart issues

.github/workflows/lint.yml

@@ -29,7 +29,7 @@ jobs:
         uses: actions/checkout@v5
 
       - name: Setup Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v6
         with:
           go-version: '~1.24'
 

.github/workflows/test.yml

@@ -28,13 +28,13 @@ jobs:
     strategy:
       matrix:
         # from https://github.com/kubernetes-sigs/controller-tools/blob/main/envtest-releases.yaml
-        envtest_k8s_version: [1.23.5, 1.33.0]
+        envtest_k8s_version: [1.23.5, 1.34.0]
     steps:
       - name: Clone the code
         uses: actions/checkout@v5
 
       - name: Setup Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v6
         with:
           go-version: '~1.24'
 

.vscode/settings.json

@@ -25,6 +25,7 @@
         "clientcmdapi",
         "clientgoscheme",
         "clientset",
+        "clientsetfake",
         "cloudnative",
         "cloudprovider",
         "clusterissuers",
@@ -46,6 +47,7 @@
         "envtest",
         "essd",
         "Eventf",
+        "featuregate",
         "finalizer",
         "Finalizers",
         "frameworkruntime",
@@ -78,6 +80,8 @@
         "iface",
         "imageutils",
         "influxdata",
+        "internalcache",
+        "internalqueue",
         "jsonpatch",
         "karpenter",
         "karpv",

api/v1/gpuresourcequota_types.go

@@ -19,7 +19,7 @@ package v1
 import (
 	v1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/kubernetes/pkg/scheduler/framework"
+	fwk "k8s.io/kube-scheduler/framework"
 )
 
 // GPUResourceQuotaSpec defines the desired state of GPUResourceQuota
@@ -188,6 +188,10 @@ type AllocRequest struct {
 	PodMeta metav1.ObjectMeta
 }
 
+func (p *AllocRequest) Clone() fwk.StateData {
+	return p
+}
+
 type GPUAllocationInfo struct {
 	Request   Resource `json:"request,omitempty"`
 	Limit     Resource `json:"limit,omitempty"`
@@ -203,7 +207,7 @@ type AdjustRequest struct {
 	NewLimit   Resource
 }
 
-func (ar *AllocRequest) Clone() framework.StateData {
+func (ar *AdjustRequest) Clone() fwk.StateData {
 	return ar
 }
 

charts/tensor-fusion/Chart.yaml

@@ -15,7 +15,7 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 1.5.6
+version: 1.5.7
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to

charts/tensor-fusion/values.yaml

@@ -31,7 +31,7 @@ controller:
   image:
     repository: tensorfusion/tensor-fusion-operator
     # Overrides the image tag whose default is the chart appVersion.
-    tag: "latest"
+    tag: "1.43.4"
   # This is for setting Kubernetes Annotations to a Pod.
   # For more information checkout: https://kubernetes.io/docs/concepts/overview/working-with-objects/annotations/ 
 
@@ -120,7 +120,7 @@ agent:
 
   image:
     repository: tensorfusion/tensor-fusion-agent
-    tag: "latest"
+    tag: "1.0.0"
 
   resources:
     requests:

cmd/main.go

@@ -27,27 +27,6 @@ import (
 
 	// Import all Kubernetes client auth plugins (e.g. Azure, GCP, OIDC, etc.)
 	// to ensure that exec-entrypoint and run can make use of them.
-
-	"k8s.io/client-go/kubernetes"
-	_ "k8s.io/client-go/plugin/pkg/client/auth"
-	"k8s.io/client-go/rest"
-	"k8s.io/klog/v2"
-
-	"k8s.io/apimachinery/pkg/runtime"
-	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
-	clientgoscheme "k8s.io/client-go/kubernetes/scheme"
-	"k8s.io/kubernetes/cmd/kube-scheduler/app"
-	"k8s.io/kubernetes/pkg/scheduler"
-	ctrl "sigs.k8s.io/controller-runtime"
-	"sigs.k8s.io/controller-runtime/pkg/client"
-	"sigs.k8s.io/controller-runtime/pkg/healthz"
-	"sigs.k8s.io/controller-runtime/pkg/manager"
-	"sigs.k8s.io/controller-runtime/pkg/metrics/filters"
-	metricsserver "sigs.k8s.io/controller-runtime/pkg/metrics/server"
-	"sigs.k8s.io/controller-runtime/pkg/webhook"
-
-	"sigs.k8s.io/yaml"
-
 	tfv1 "github.com/NexusGPU/tensor-fusion/api/v1"
 	"github.com/NexusGPU/tensor-fusion/cmd/sched"
 	"github.com/NexusGPU/tensor-fusion/internal/alert"
@@ -65,6 +44,25 @@ import (
 	"github.com/NexusGPU/tensor-fusion/internal/utils"
 	"github.com/NexusGPU/tensor-fusion/internal/version"
 	webhookcorev1 "github.com/NexusGPU/tensor-fusion/internal/webhook/v1"
+	"k8s.io/apimachinery/pkg/runtime"
+	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
+	k8sVer "k8s.io/apimachinery/pkg/util/version"
+	"k8s.io/apiserver/pkg/util/feature"
+	"k8s.io/client-go/kubernetes"
+	clientgoscheme "k8s.io/client-go/kubernetes/scheme"
+	_ "k8s.io/client-go/plugin/pkg/client/auth"
+	"k8s.io/client-go/rest"
+	"k8s.io/klog/v2"
+	"k8s.io/kubernetes/cmd/kube-scheduler/app"
+	"k8s.io/kubernetes/pkg/scheduler"
+	ctrl "sigs.k8s.io/controller-runtime"
+	"sigs.k8s.io/controller-runtime/pkg/client"
+	"sigs.k8s.io/controller-runtime/pkg/healthz"
+	"sigs.k8s.io/controller-runtime/pkg/manager"
+	"sigs.k8s.io/controller-runtime/pkg/metrics/filters"
+	metricsserver "sigs.k8s.io/controller-runtime/pkg/metrics/server"
+	"sigs.k8s.io/controller-runtime/pkg/webhook"
+	"sigs.k8s.io/yaml"
 	// +kubebuilder:scaffold:imports
 )
 
@@ -204,6 +202,14 @@ func main() {
 	_ = os.Setenv(constants.KubeApiVersionMajorEnv, version.Major)
 	_ = os.Setenv(constants.KubeApiVersionMinorEnv, version.Minor)
 
+	// TODO: there will still be risk after FeatureGate removed when the feature is stable for a long time
+	// To be compatible with long-term k8s version, need to patch Kubernetes source code
+	k8sVersion := k8sVer.MustParseSemantic(version.String())
+	err = feature.DefaultMutableFeatureGate.SetEmulationVersion(k8sVersion)
+	if err != nil {
+		setupLog.Error(err, "unable to set k8s version for feature gating")
+	}
+
 	alertEvaluatorReady = make(chan struct{})
 	setupTimeSeriesAndWatchGlobalConfigChanges(ctx, mgr)
 
@@ -221,7 +227,7 @@ func main() {
 	pricingProvider := pricing.NewStaticPricingProvider()
 	startWebhook(mgr, portAllocator, pricingProvider)
 
-	scheduler := startScheduler(ctx, allocator, mgr)
+	scheduler := startScheduler(ctx, allocator, mgr, k8sVersion)
 
 	startCustomResourceController(ctx, mgr, metricsRecorder, allocator, portAllocator)
 
@@ -461,6 +467,7 @@ func startScheduler(
 	ctx context.Context,
 	allocator *gpuallocator.GpuAllocator,
 	mgr manager.Manager,
+	k8sVersion *k8sVer.Version,
 ) *scheduler.Scheduler {
 	if os.Getenv(constants.EnableSchedulerEnv) == constants.FalseStringValue {
 		return nil
@@ -479,7 +486,9 @@ func startScheduler(
 		gpuTopoPlugin.NewWithDeps(allocator, mgr.GetClient()),
 	)
 
-	cc, scheduler, err := sched.SetupScheduler(ctx, mgr, schedulerConfigPath, false, gpuResourceFitOpt, gpuTopoOpt)
+	cc, scheduler, err := sched.SetupScheduler(
+		ctx, mgr, schedulerConfigPath, false, k8sVersion, gpuResourceFitOpt, gpuTopoOpt,
+	)
 	if err != nil {
 		setupLog.Error(err, "unable to create tensor fusion scheduler")
 		os.Exit(1)

cmd/sched/setup.go

@@ -22,6 +22,8 @@ import (
 	"strings"
 
 	utilerrors "k8s.io/apimachinery/pkg/util/errors"
+	k8sVer "k8s.io/apimachinery/pkg/util/version"
+	"k8s.io/apiserver/pkg/util/feature"
 	"k8s.io/client-go/tools/events"
 	"k8s.io/component-base/configz"
 	"k8s.io/klog/v2"
@@ -50,6 +52,7 @@ func SetupScheduler(
 	mgr manager.Manager,
 	schedulerConfigPath string,
 	disableHttpEndpoint bool,
+	k8sVersion *k8sVer.Version,
 	outOfTreeRegistryOptions ...app.Option,
 ) (*schedulerserverconfig.CompletedConfig, *scheduler.Scheduler, error) {
 	opts := options.NewOptions()
@@ -73,6 +76,12 @@ func SetupScheduler(
 		return nil, nil, err
 	}
 
+	// Setup enumerationVersion again since it's overridden by the config
+	err = feature.DefaultMutableFeatureGate.SetEmulationVersion(k8sVersion)
+	if err != nil {
+		return nil, nil, err
+	}
+
 	if cfg, err := latest.Default(); err != nil {
 		return nil, nil, err
 	} else {