Add first e2e tests to the repo

.gitignore

@@ -1,4 +1,5 @@
 bin/
 .idea/
 charts/
-cover.out
+cover.out
+.DS_Store

build/makefile/testenv.mk

@@ -2,6 +2,9 @@
 ENVTEST_K8S_VERSION = 1.32.0
 ENVTEST_VERSION=release-0.20
 
+E2E_TESTS_DIR = "test/e2e/"
+TEST_TARGETS = $(shell go list ./... | grep -v "${E2E_TESTS_DIR}")
+
 envtest-docker-go: gocache
 	@ ${ECHO_COMMAND} ${GREEN_CONSOLE} "${CONSOLE_PREFIX} Running unit-tests" ${BASE_CONSOLE}
 	${DOCKER_GO_COMMAND} make envtest-go || ${FAILURE_MESSAGE_HANDLER}
@@ -10,7 +13,7 @@ envtest-docker-go: gocache
 envtest-go: envtest
 	@ ${ECHO_COMMAND} ${GREEN_CONSOLE} "${CONSOLE_PREFIX} Running unit-tests" ${BASE_CONSOLE}
 	KUBEBUILDER_ASSETS="$(shell $(ENVTEST) use $(ENVTEST_K8S_VERSION) -p path --bin-dir $(LOCALBIN))" \
-    go test ./... -timeout 30m || ${FAILURE_MESSAGE_HANDLER}
+    go test ${TEST_TARGETS} -timeout 30m || ${FAILURE_MESSAGE_HANDLER}
 	${SUCCESS_MESSAGE_HANDLER}
 
 ENVTEST = $(LOCALBIN)/setup-envtest

go.mod

@@ -3,17 +3,20 @@ module github.com/NVIDIA/KAI-scheduler
 go 1.23.4
 
 require (
+	github.com/NVIDIA/gpu-operator v1.8.3-0.20240812232433-87286e93f2c9
 	github.com/argoproj/argo-workflows/v3 v3.6.4
 	github.com/dustin/go-humanize v1.0.1
 	github.com/gin-contrib/pprof v1.5.2
 	github.com/gin-gonic/gin v1.10.0
+	github.com/go-logr/logr v1.4.2
 	github.com/golang/glog v1.2.4
 	github.com/grafana/pyroscope-go v1.2.1
 	github.com/onsi/ginkgo v1.16.5
 	github.com/onsi/ginkgo/v2 v2.22.2
 	github.com/onsi/gomega v1.36.2
 	github.com/pkg/errors v0.9.1
 	github.com/prometheus/client_golang v1.20.5
+	github.com/run-ai/kwok-operator v0.0.0-20240926063032-05b6364bc7c7
 	github.com/spf13/pflag v1.0.6
 	github.com/stretchr/testify v1.10.0
 	github.com/xyproto/randomstring v1.2.0
@@ -51,6 +54,7 @@ require (
 	k8s.io/pod-security-admission v0.32.1
 	k8s.io/sample-apiserver v0.32.1
 	k8s.io/utils v0.0.0-20241104100929-3ea5e8cea738
+	knative.dev/pkg v0.0.0-20250117084104-c43477f0052b
 	knative.dev/serving v0.44.0
 	sigs.k8s.io/controller-runtime v0.20.0
 	sigs.k8s.io/karpenter v1.2.0
@@ -59,9 +63,11 @@ require (
 require (
 	cel.dev/expr v0.18.0 // indirect
 	github.com/Microsoft/go-winio v0.6.2 // indirect
+	github.com/NVIDIA/k8s-kata-manager v0.2.0 // indirect
+	github.com/NVIDIA/k8s-operator-libs v0.0.0-20240627150410-078e3039ecf7 // indirect
 	github.com/NYTimes/gziphandler v1.1.1 // indirect
 	github.com/antlr4-go/antlr/v4 v4.13.0 // indirect
-	github.com/asaskevich/govalidator v0.0.0-20190424111038-f61b66f89f4a // indirect
+	github.com/asaskevich/govalidator v0.0.0-20230301143203-a9d515a09cc2 // indirect
 	github.com/awslabs/operatorpkg v0.0.0-20241205163410-0fff9f28d115 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
 	github.com/blang/semver/v4 v4.0.0 // indirect
@@ -83,7 +89,6 @@ require (
 	github.com/fxamacker/cbor/v2 v2.7.0 // indirect
 	github.com/gabriel-vasile/mimetype v1.4.7 // indirect
 	github.com/gin-contrib/sse v0.1.0 // indirect
-	github.com/go-logr/logr v1.4.2 // indirect
 	github.com/go-logr/stdr v1.2.2 // indirect
 	github.com/go-logr/zapr v1.3.0 // indirect
 	github.com/go-openapi/jsonpointer v0.21.0 // indirect
@@ -132,6 +137,7 @@ require (
 	github.com/opencontainers/selinux v1.11.1 // indirect
 	github.com/pelletier/go-toml/v2 v2.2.3 // indirect
 	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
+	github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.73.2 // indirect
 	github.com/prometheus/client_model v0.6.1 // indirect
 	github.com/prometheus/common v0.55.0 // indirect
 	github.com/prometheus/procfs v0.15.1 // indirect
@@ -157,6 +163,7 @@ require (
 	go.opentelemetry.io/proto/otlp v1.3.1 // indirect
 	golang.org/x/arch v0.12.0 // indirect
 	golang.org/x/crypto v0.35.0 // indirect
+	golang.org/x/mod v0.22.0 // indirect
 	golang.org/x/net v0.36.0 // indirect
 	golang.org/x/oauth2 v0.25.0 // indirect
 	golang.org/x/sync v0.11.0 // indirect
@@ -183,7 +190,6 @@ require (
 	k8s.io/kube-scheduler v0.32.1 // indirect
 	k8s.io/kubelet v0.32.1 // indirect
 	knative.dev/networking v0.0.0-20250117155906-67d1c274ba6a // indirect
-	knative.dev/pkg v0.0.0-20250117084104-c43477f0052b // indirect
 	sigs.k8s.io/apiserver-network-proxy/konnectivity-client v0.31.0 // indirect
 	sigs.k8s.io/json v0.0.0-20241010143419-9aa6b5e7a4b3 // indirect
 	sigs.k8s.io/structured-merge-diff/v4 v4.4.2 // indirect

go.sum

@@ -9,6 +9,12 @@ contrib.go.opencensus.io/exporter/prometheus v0.4.2/go.mod h1:dvEHbiKmgvbr5pjaF9
 github.com/BurntSushi/toml v0.3.1/go.mod h1:xHWCNGjB5oqiDr8zfno3MHue2Ht5sIBksp03qcyfWMU=
 github.com/Microsoft/go-winio v0.6.2 h1:F2VQgta7ecxGYO8k3ZZz3RS8fVIXVxONVUPlNERoyfY=
 github.com/Microsoft/go-winio v0.6.2/go.mod h1:yd8OoFMLzJbo9gZq8j5qaps8bJ9aShtEA8Ipt1oGCvU=
+github.com/NVIDIA/gpu-operator v1.8.3-0.20240812232433-87286e93f2c9 h1:9yw3Jkto9ZqtNwlnOzxAlKufMNplaWPRfkXhCUFHXgI=
+github.com/NVIDIA/gpu-operator v1.8.3-0.20240812232433-87286e93f2c9/go.mod h1:lOgoRYbt1dtCVGX+EhxuZgPonfcIs41BXrnIPk3fE3I=
+github.com/NVIDIA/k8s-kata-manager v0.2.0 h1:K+BFkXTOvXXj/kmbNfxFCXM+GkdOVZj2WTHQ7b2uQA0=
+github.com/NVIDIA/k8s-kata-manager v0.2.0/go.mod h1:fVUz0DLzwW9RQBE59cLNTi3LrzVwSXWIogr9y5FocPM=
+github.com/NVIDIA/k8s-operator-libs v0.0.0-20240627150410-078e3039ecf7 h1:NaClubDuTKoXy4Ev4ZkmpVy3u6xdwd3I1XFJdoI1r+M=
+github.com/NVIDIA/k8s-operator-libs v0.0.0-20240627150410-078e3039ecf7/go.mod h1:d8YV6Am03Z9VS4fh6virN/ltOSasCZtAhkwMRU1X6Vs=
 github.com/NYTimes/gziphandler v1.1.1 h1:ZUDjpQae29j0ryrS0u/B8HZfJBtBQHjqw2rQ2cqUQ3I=
 github.com/NYTimes/gziphandler v1.1.1/go.mod h1:n/CVRwUEOgIxrgPvAQhUUr9oeUtvrhMomdKFjzJNB0c=
 github.com/Pallinder/go-randomdata v1.2.0 h1:DZ41wBchNRb/0GfsePLiSwb0PHZmT67XY00lCDlaYPg=
@@ -20,8 +26,8 @@ github.com/argoproj/argo-workflows/v3 v3.6.4 h1:5+Cc1UwaQE5ka3w7R3hxZ1TK3M6VjDEX
 github.com/argoproj/argo-workflows/v3 v3.6.4/go.mod h1:2f5zB8CkbNCCO1od+kd1dWkVokqcuyvu+tc+Jwx1MZg=
 github.com/armon/go-socks5 v0.0.0-20160902184237-e75332964ef5 h1:0CwZNZbxp69SHPdPJAN/hZIm0C4OItdklCFmMRWYpio=
 github.com/armon/go-socks5 v0.0.0-20160902184237-e75332964ef5/go.mod h1:wHh0iHkYZB8zMSxRWpUBQtwG5a7fFgvEO+odwuTv2gs=
-github.com/asaskevich/govalidator v0.0.0-20190424111038-f61b66f89f4a h1:idn718Q4B6AGu/h5Sxe66HYVdqdGu2l9Iebqhi/AEoA=
-github.com/asaskevich/govalidator v0.0.0-20190424111038-f61b66f89f4a/go.mod h1:lB+ZfQJz7igIIfQNfa7Ml4HSf2uFQQRzpGGRXenZAgY=
+github.com/asaskevich/govalidator v0.0.0-20230301143203-a9d515a09cc2 h1:DklsrG3dyBCFEj5IhUbnKptjxatkF07cF2ak3yi77so=
+github.com/asaskevich/govalidator v0.0.0-20230301143203-a9d515a09cc2/go.mod h1:WaHUgvxTVq04UNunO+XhnAqY/wQc+bxr74GqbsZ/Jqw=
 github.com/avast/retry-go v3.0.0+incompatible h1:4SOWQ7Qs+oroOTQOYnAHqelpCO0biHSxpiH9JdtuBj0=
 github.com/avast/retry-go v3.0.0+incompatible/go.mod h1:XtSnn+n/sHqQIpZ10K1qAevBhOOCWBLXXy3hyiqqBrY=
 github.com/awslabs/operatorpkg v0.0.0-20241205163410-0fff9f28d115 h1:9nhjY3dzCpEmhpQ0vMlhB7wqucAiftLjAIEQu8uT2J4=
@@ -258,6 +264,8 @@ github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINE
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.73.2 h1:GwlGJPK6vf1UIohpc72KJVkKYlzki1UgE3xC4bWbf20=
+github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.73.2/go.mod h1:yJ3CawR/A5qEYFEeCOUVYLTwYxmacfHQhJS+b/2QiaM=
 github.com/prometheus/client_golang v1.20.5 h1:cxppBPuYhUnsO6yo/aoRol4L7q7UFfdm+bR9r+8l63Y=
 github.com/prometheus/client_golang v1.20.5/go.mod h1:PIEt8X02hGcP8JWbeHyeZ53Y/jReSnHgO035n//V5WE=
 github.com/prometheus/client_model v0.0.0-20190812154241-14fe0d1b01d4/go.mod h1:xMI15A0UPsDsEKsMN9yxemIoYk6Tm2C1GtYGdfGttqA=
@@ -274,6 +282,8 @@ github.com/robfig/cron/v3 v3.0.1/go.mod h1:eQICP3HwyT7UooqI/z+Ov+PtYAWygg1TEWWzG
 github.com/rogpeppe/fastuuid v1.2.0/go.mod h1:jVj6XXZzXRy/MSR5jhDC/2q6DgLz+nrA6LYCDYWNEvQ=
 github.com/rogpeppe/go-internal v1.12.0 h1:exVL4IDcn6na9z1rAb56Vxr+CgyK3nn3O+epU5NdKM8=
 github.com/rogpeppe/go-internal v1.12.0/go.mod h1:E+RYuTGaKKdloAfM02xzb0FW3Paa99yedzYV+kq4uf4=
+github.com/run-ai/kwok-operator v0.0.0-20240926063032-05b6364bc7c7 h1:7sUlviMShcd8g6sf2Q93ix+geV0staOMk42Rs0rAJqA=
+github.com/run-ai/kwok-operator v0.0.0-20240926063032-05b6364bc7c7/go.mod h1:vih5aAo7hS8Mt6NXsIUI8SYW+WfB4GhZOuLZzLjZWcc=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
 github.com/samber/lo v1.47.0 h1:z7RynLwP5nbyRscyvcD043DWYoOcYRv3mV8lBeqOCLc=
 github.com/samber/lo v1.47.0/go.mod h1:RmDH9Ct32Qy3gduHQuKJ3gW1fMHAnE/fAzQuf6He5cU=

hack/e2e-kind-config.yaml

@@ -0,0 +1,16 @@
+# Copyright 2025 NVIDIA CORPORATION
+# SPDX-License-Identifier: Apache-2.0
+
+kind: Cluster
+apiVersion: kind.x-k8s.io/v1alpha4
+nodes:
+- role: control-plane
+- role: worker
+  labels:
+    run.ai/simulated-gpu-node-pool: default
+- role: worker
+  labels:
+    run.ai/simulated-gpu-node-pool: default
+- role: worker
+  labels:
+    run.ai/simulated-gpu-node-pool: default

hack/run-e2e-kind.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+# Copyright 2025 NVIDIA CORPORATION
+# SPDX-License-Identifier: Apache-2.0
+
+
+CLUSTER_NAME=${CLUSTER_NAME:-e2e-kai-scheduler}
+
+REPO_ROOT=$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )/..
+KIND_CONFIG=${REPO_ROOT}/hack/e2e-kind-config.yaml
+GOPATH=${HOME}/go
+GOBIN=${GOPATH}/bin
+
+NVCR_SECRET_FILE_PATH=${1}
+if [ -z "$NVCR_SECRET_FILE_PATH" ]; then
+    echo "Must a path to an appropriate secret file, that contains the credentials for the nvstaging-runai helm and docker image repository"
+    exit 1
+fi
+
+kind create cluster --config ${KIND_CONFIG} --name $CLUSTER_NAME
+
+kubectl create namespace kai-scheduler
+# Set an appropriate secret to allow the kube-ai system pods to pull from nvstaging-runai and pull test images for the e2e tests
+kubectl apply -f ${NVCR_SECRET_FILE_PATH} -n kai-scheduler
+
+
+# Install the fake-gpu-operator to provide a fake GPU resources for the e2e tests
+helm upgrade -i gpu-operator fake-gpu-operator/fake-gpu-operator --namespace gpu-operator --create-namespace --version 0.0.53 --set topology.nodePools.default.gpuCount=8
+
+helm upgrade -i kai-scheduler nvstaging-runai/kai-scheduler -n kai-scheduler --create-namespace --set "global.imagePullSecrets[0].name=nvcr-secret" --set "global.gpuSharing=true" --set "global.registry=nvcr.io/nvstaging/runai" --version v0.2.0
+
+# Allow all the pods in the fake-gpu-operator and kai-scheduler to start
+sleep 30
+
+# Install ginkgo if it's not installed
+if [ ! -f ${GOBIN}/ginkgo ]; then
+    echo "Installing ginkgo"
+    GOBIN=${GOBIN} go install github.com/onsi/ginkgo/v2/[email protected]
+fi
+
+${GOBIN}/ginkgo -r --keep-going --randomize-all --randomize-suites --trace -vv ${REPO_ROOT}/test/e2e/suites --label-filter '!autoscale', '!scale'
+
+kind delete cluster --name $CLUSTER_NAME

pkg/common/constants/constants.go

@@ -33,4 +33,9 @@ const (
 	MigEnabledLabel          = "node-role.kubernetes.io/runai-mig-enabled"
 	MigStrategyLabel         = "nvidia.com/mig.strategy"
 	GpuCountLabel            = "nvidia.com/gpu.count"
+	QueueLabelKey            = "runai/queue"
+
+	// Namespaces
+	SystemPodsNamespace       = "kai-scheduler"
+	RunaiReservationNamespace = "runai-reservation"
 )

pkg/common/gpu_operator_discovery/cdi.go

@@ -0,0 +1,46 @@
+// Copyright 2025 NVIDIA CORPORATION
+// SPDX-License-Identifier: Apache-2.0
+
+package gpu_operator_discovery
+
+import (
+	"context"
+	"fmt"
+
+	nvidiav1 "github.com/NVIDIA/gpu-operator/api/nvidia/v1"
+	kerrors "k8s.io/apimachinery/pkg/api/errors"
+	"k8s.io/apimachinery/pkg/api/meta"
+	"sigs.k8s.io/controller-runtime/pkg/client"
+	logf "sigs.k8s.io/controller-runtime/pkg/log"
+)
+
+func IsCdiEnabled(ctx context.Context, readerClient client.Reader) (bool, error) {
+	nvidiaClusterPolicies := &nvidiav1.ClusterPolicyList{}
+	err := readerClient.List(ctx, nvidiaClusterPolicies)
+	if err != nil {
+		if meta.IsNoMatchError(err) || kerrors.IsNotFound(err) {
+			return false, nil
+		}
+		log := logf.FromContext(ctx)
+		log.Error(err, "cannot list nvidia cluster policy")
+		return false, err
+	}
+
+	if len(nvidiaClusterPolicies.Items) == 0 {
+		return false, nil
+	}
+	if len(nvidiaClusterPolicies.Items) > 1 {
+		log := logf.FromContext(ctx)
+		log.Info(fmt.Sprintf("Cluster has %d clusterpolicies.nvidia.com/v1 objects."+
+			" First one is queried for the cdi configuration", len(nvidiaClusterPolicies.Items)))
+	}
+
+	nvidiaClusterPolicy := nvidiaClusterPolicies.Items[0]
+	if nvidiaClusterPolicy.Spec.CDI.Enabled != nil && *nvidiaClusterPolicy.Spec.CDI.Enabled {
+		if nvidiaClusterPolicy.Spec.CDI.Default != nil && *nvidiaClusterPolicy.Spec.CDI.Default {
+			return true, nil
+		}
+	}
+
+	return false, nil
+}

test/e2e/modules/constant/constant.go

@@ -0,0 +1,17 @@
+/*
+Copyright 2025 NVIDIA CORPORATION
+SPDX-License-Identifier: Apache-2.0
+*/
+
+package constant
+
+const (
+	NvidiaGPUMemoryLabelName        = "nvidia.com/gpu.memory"
+	NodeNamePodLabelName            = "kubernetes.io/hostname"
+	RunaiSchedulerName              = "kai-scheduler"
+	RunaiReservationNamespace       = "runai-reservation"
+	SystemPodsNamespace             = "kai-scheduler"
+	NonPreemptiblePriorityThreshold = 100
+	EngineTestPodsApp               = "engine-e2e"
+	QueueLabelKey                   = "runai/queue"
+)

test/e2e/modules/constant/labels/labels.go

@@ -0,0 +1,9 @@
+/*
+Copyright 2025 NVIDIA CORPORATION
+SPDX-License-Identifier: Apache-2.0
+*/
+package labels
+
+const (
+	ReservationPod = "reservationPod"
+)

test/e2e/modules/context/cluster.go

@@ -0,0 +1,51 @@
+/*
+Copyright 2025 NVIDIA CORPORATION
+SPDX-License-Identifier: Apache-2.0
+*/
+package context
+
+import (
+	"context"
+
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+
+	v2 "github.com/NVIDIA/KAI-scheduler/pkg/apis/scheduling/v2"
+	"github.com/NVIDIA/KAI-scheduler/test/e2e/modules/resources/rd"
+	"github.com/NVIDIA/KAI-scheduler/test/e2e/modules/resources/rd/queue"
+)
+
+const (
+	defaultServiceAccountName = "default"
+)
+
+func (tc *TestContext) createClusterQueues(ctx context.Context) error {
+	for _, testQueue := range tc.Queues {
+		err := createQueueContext(ctx, testQueue)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func createQueueContext(ctx context.Context, q *v2.Queue) error {
+	_, err := queue.Create(kubeAiSchedClientset, ctx, q, metav1.CreateOptions{})
+	if err != nil {
+		return err
+	}
+
+	namespaceName := queue.GetConnectedNamespaceToQueue(q)
+	ns := rd.CreateNamespaceObject(namespaceName, q.Name)
+	_, err = kubeClientset.
+		CoreV1().
+		Namespaces().
+		Create(ctx, ns, metav1.CreateOptions{})
+	if err != nil {
+		return err
+	}
+
+	// TODO: add RBAC role bindings
+	// TODO: patch the namespace to add appropriate secret to the service account
+
+	return nil
+}