e2e: Optimize installation and replacement of kdoctor

Signed-off-by: ty-dc <[email protected]>

Makefile

@@ -308,7 +308,7 @@ e2e_init:
 
 .PHONY: e2e_init_underlay
 e2e_init_underlay:
-	$(QUIET)  make e2e_init -e INSTALL_OVERLAY_CNI=false -e E2E_SPIDERPOOL_ENABLE_SUBNET=true
+	$(QUIET)  make e2e_init -e INSTALL_OVERLAY_CNI=false -e E2E_SPIDERPOOL_ENABLE_SUBNET=true -e INSTALL_KRUISE=true
 
 .PHONY: e2e_init_cilium_with_ebpf
 e2e_init_cilium_with_ebpf:

go.mod

@@ -15,7 +15,7 @@ require (
 	github.com/go-openapi/strfmt v0.21.7
 	github.com/go-openapi/swag v0.22.4
 	github.com/go-openapi/validate v0.22.1
-	github.com/go-swagger/go-swagger v0.30.3
+	github.com/go-swagger/go-swagger v0.30.4
 	github.com/gogo/protobuf v1.3.2
 	github.com/golang/mock v1.6.0
 	github.com/google/gops v0.3.27
@@ -34,7 +34,6 @@ require (
 	github.com/spf13/cobra v1.7.0
 	github.com/spf13/pflag v1.0.5
 	github.com/spidernet-io/e2eframework v0.0.0-20230724150324-2eee77078275
-	github.com/spidernet-io/spiderdoctor v0.3.0
 	github.com/tigera/operator v1.30.5
 	github.com/vishvananda/netlink v1.2.1-beta.2.0.20230621221334-77712cff8739
 	go.opentelemetry.io/otel v1.20.0
@@ -46,7 +45,7 @@ require (
 	go.uber.org/atomic v1.10.0
 	go.uber.org/multierr v1.11.0
 	go.uber.org/zap v1.25.0
-	golang.org/x/net v0.13.0
+	golang.org/x/net v0.17.0
 	golang.org/x/sync v0.3.0
 	golang.org/x/sys v0.14.0
 	golang.org/x/tools v0.11.0
@@ -71,10 +70,12 @@ require (
 	k8s.io/klog/v2 v2.100.1 // indirect
 )
 
+require github.com/kdoctor-io/kdoctor v0.2.0
+
 require (
 	github.com/Masterminds/goutils v1.1.1 // indirect
-	github.com/Masterminds/semver/v3 v3.1.1 // indirect
-	github.com/Masterminds/sprig/v3 v3.2.2 // indirect
+	github.com/Masterminds/semver/v3 v3.2.0 // indirect
+	github.com/Masterminds/sprig/v3 v3.2.3 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
 	github.com/blang/semver/v4 v4.0.0 // indirect
 	github.com/cespare/xxhash/v2 v2.2.0 // indirect
@@ -107,7 +108,7 @@ require (
 	github.com/gorilla/handlers v1.5.1 // indirect
 	github.com/grafana/pyroscope-go/godeltaprof v0.1.3 // indirect
 	github.com/hashicorp/hcl v1.0.0 // indirect
-	github.com/huandu/xstrings v1.3.2 // indirect
+	github.com/huandu/xstrings v1.3.3 // indirect
 	github.com/imdario/mergo v0.3.13 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/josharian/intern v1.0.0 // indirect
@@ -163,12 +164,12 @@ require (
 	github.com/yusufpapurcu/wmi v1.2.3 // indirect
 	go.mongodb.org/mongo-driver v1.11.3 // indirect
 	go.uber.org/dig v1.17.0 // indirect
-	golang.org/x/crypto v0.11.0 // indirect
+	golang.org/x/crypto v0.14.0 // indirect
 	golang.org/x/exp v0.0.0-20230522175609-2e198f4a06a1 // indirect
 	golang.org/x/mod v0.12.0 // indirect
-	golang.org/x/oauth2 v0.8.0 // indirect
-	golang.org/x/term v0.10.0 // indirect
-	golang.org/x/text v0.11.0 // indirect
+	golang.org/x/oauth2 v0.10.0 // indirect
+	golang.org/x/term v0.13.0 // indirect
+	golang.org/x/text v0.13.0 // indirect
 	golang.org/x/time v0.3.0 // indirect
 	gomodules.xyz/jsonpatch/v2 v2.4.0 // indirect
 	google.golang.org/appengine v1.6.7 // indirect

test/Makefile

@@ -19,7 +19,6 @@ kind-init: check_env prepare
 	E2E_IP_FAMILY=$(E2E_IP_FAMILY) \
 		E2E_CLUSTER_NAME=$(E2E_CLUSTER_NAME)  \
 		E2E_KUBECONFIG=$(E2E_KUBECONFIG)  \
-		INSTALL_NETTOOLS=$(INSTALL_NETTOOLS) \
 		E2E_VLAN_GATEWAY_IMAGE=$(E2E_VLAN_GATEWAY_IMAGE) \
 		VLAN_GATEWAY_CONTAINER=$(VLAN_GATEWAY_CONTAINER) \
 		$(QUIET) bash scripts/config-network.sh
@@ -47,7 +46,7 @@ endif
 	@echo -e "\033[35m [Step 7] Install Spiderpool \033[0m"
 	@ make setup_spiderpool
 ifeq ($(INSTALL_MULTUS),true)
-	@echo -e "\033[35m [Step 8] Install Multus \033[0m"
+	@echo -e "\033[35m [Step 8] Install Multus CR and IPPool CR \033[0m"
 	@ IMAGE_MULTUS=$(IMAGE_MULTUS_NAME) \
 		E2E_IP_FAMILY=$(E2E_IP_FAMILY) \
 		MULTUS_DEFAULT_CNI_NAME=$(MULTUS_DEFAULT_CNI_VLAN0) \
@@ -70,16 +69,16 @@ ifeq ($(INSTALL_KRUISE),true)
 	@echo -e "\033[35m [Step 9] Install third-party controllers: kurise \033[0m"
 	@ make setup_kurise
 endif
-ifeq ($(INSTALL_SPIDERDOCTOR),true)
-	@echo -e "\033[35m [Step 10] Install SpiderDoctor \033[0m"
+ifeq ($(INSTALL_KDOCTOR),true)
+	@echo -e "\033[35m [Step 10] Install kdoctor \033[0m"
 	E2E_KUBECONFIG=$(E2E_KUBECONFIG)  \
 		E2E_IP_FAMILY=$(E2E_IP_FAMILY) \
 		E2E_CLUSTER_NAME=$(E2E_CLUSTER_NAME) \
-		SPIDERDOCTOR_VERSION=$(SPIDERDOCTOR_VERSION) \
-		SPIDERDOCTOR_REPORT_PATH=$(SPIDERDOCTOR_REPORT_PATH) \
-		E2E_SPIDERDOCTOR_IMAGE_REPO=$(E2E_SPIDERDOCTOR_IMAGE_REPO) \
+		KDOCTOR_VERSION=$(KDOCTOR_VERSION) \
+		KDOCTOR_REPORT_PATH=$(KDOCTOR_REPORT_PATH) \
+		E2E_KDOCTOR_IMAGE_REPO=$(E2E_KDOCTOR_IMAGE_REPO) \
 		HTTP_PROXY=$(HTTP_PROXY) \
-		$(QUIET) bash scripts/install-spiderdoctor.sh
+		$(QUIET) bash scripts/install-kdoctor.sh
 endif
 ifeq ($(INSTALL_KUBEVIRT),true)
 	@echo -e "\033[35m [Step 10] Install kubevirt \033[0m"

test/Makefile.defs

@@ -41,7 +41,7 @@ INSTALL_MACVLAN ?= true
 
 INSTALL_MULTUS ?= true
 
-INSTALL_KRUISE ?= true
+INSTALL_KRUISE ?= false
 
 INSTALL_OVERLAY_CNI ?= false
 
@@ -51,9 +51,7 @@ INSTALL_CILIUM ?= true
 
 DISABLE_KUBE_PROXY ?= false
 
-INSTALL_NETTOOLS ?= false
-
-INSTALL_SPIDERDOCTOR ?= true
+INSTALL_KDOCTOR ?= true
 
 INSTALL_KUBEVIRT ?= false
 
@@ -150,16 +148,16 @@ endif
 
 CILIUM_VERSION ?=
 
-#================= spiderdoctor
-SPIDERDOCTOR_VERSION ?= v0.3.0
+#================= kdoctor
+KDOCTOR_VERSION ?= v0.2.0
 
 ifeq ($(E2E_CHINA_IMAGE_REGISTRY),true)
-    E2E_SPIDERDOCTOR_IMAGE_REPO ?= ghcr.m.daocloud.io
+    E2E_KDOCTOR_IMAGE_REPO ?= ghcr.m.daocloud.io
 else
-    E2E_SPIDERDOCTOR_IMAGE_REPO ?= ghcr.io
+    E2E_KDOCTOR_IMAGE_REPO ?= ghcr.io
 endif
 
-SPIDERDOCTOR_REPORT_PATH=/var/log/spiderdoctor/controller
+KDOCTOR_REPORT_PATH=/var/log/kdoctor/controller
 
 #====== pyroscope
 PYROSCOPE_CONTAINER_NAME := e2e-pyroscope

test/e2e/common/constant.go

@@ -72,9 +72,9 @@ var (
 	SpiderPoolIPv4SubnetVlan200 string = "vlan200-v4"
 	SpiderPoolIPv6SubnetVlan200 string = "vlan200-v6"
 
-	MultusNs                = "kube-system"
-	SpiderDoctorAgentNs     = "kube-system"
-	SpiderDoctorAgentDSName = "spiderdoctor-agent"
+	MultusNs           = "kube-system"
+	KDoctorAgentNs     = "kube-system"
+	KDoctorAgentDSName = "kdoctor-agent"
 
 	// gateway and check for ip conflicting machines
 	VlanGatewayContainer = "vlan-gateway"

test/e2e/coordinator/macvlan-overlay-one/macvlan_overlay_one_suite_test.go

@@ -6,10 +6,10 @@ import (
 	"testing"
 
 	multus_v1 "github.com/k8snetworkplumbingwg/network-attachment-definition-client/pkg/apis/k8s.cni.cncf.io/v1"
+	kdoctorV1beta1 "github.com/kdoctor-io/kdoctor/pkg/k8s/apis/kdoctor.io/v1beta1"
 	. "github.com/onsi/ginkgo/v2"
 	. "github.com/onsi/gomega"
 	e2e "github.com/spidernet-io/e2eframework/framework"
-	spiderdoctorV1 "github.com/spidernet-io/spiderdoctor/pkg/k8s/apis/spiderdoctor.spidernet.io/v1"
 	spiderpool "github.com/spidernet-io/spiderpool/pkg/k8s/apis/spiderpool.spidernet.io/v2beta1"
 	"github.com/spidernet-io/spiderpool/test/e2e/common"
 	appsv1 "k8s.io/api/apps/v1"
@@ -24,24 +24,24 @@ func TestMacvlanOverlayOne(t *testing.T) {
 var frame *e2e.Framework
 
 // var name string
-var spiderDoctorAgent *appsv1.DaemonSet
+var kdoctorAgent *appsv1.DaemonSet
 
 var successRate = float64(1)
 var name string
 var err error
 var delayMs = int64(15000)
 var (
-	task        *spiderdoctorV1.Nethttp
-	plan        *spiderdoctorV1.SchedulePlan
-	target      *spiderdoctorV1.NethttpTarget
-	targetAgent *spiderdoctorV1.TargetAgentSepc
-	request     *spiderdoctorV1.NethttpRequest
-	condition   *spiderdoctorV1.NetSuccessCondition
+	task        *kdoctorV1beta1.NetReach
+	plan        *kdoctorV1beta1.SchedulePlan
+	targetAgent *kdoctorV1beta1.NetReachTarget
+	request     *kdoctorV1beta1.NetHttpRequest
+	schedule    *kdoctorV1beta1.SchedulePlan
+	condition   *kdoctorV1beta1.NetSuccessCondition
 	run         = true
 )
 
 var _ = BeforeSuite(func() {
-	frame, err = e2e.NewFramework(GinkgoT(), []func(*runtime.Scheme) error{multus_v1.AddToScheme, spiderpool.AddToScheme, spiderdoctorV1.AddToScheme})
+	frame, err = e2e.NewFramework(GinkgoT(), []func(*runtime.Scheme) error{multus_v1.AddToScheme, spiderpool.AddToScheme, kdoctorV1beta1.AddToScheme})
 	Expect(err).NotTo(HaveOccurred())
 
 	if !common.CheckRunOverlayCNI() {

test/e2e/coordinator/macvlan-overlay-one/macvlan_overlay_one_test.go

@@ -15,7 +15,7 @@ import (
 
 	"k8s.io/utils/pointer"
 
-	spiderdoctorV1 "github.com/spidernet-io/spiderdoctor/pkg/k8s/apis/spiderdoctor.spidernet.io/v1"
+	kdoctorV1beta1 "github.com/kdoctor-io/kdoctor/pkg/k8s/apis/kdoctor.io/v1beta1"
 	"github.com/spidernet-io/spiderpool/pkg/constant"
 	"github.com/spidernet-io/spiderpool/pkg/ip"
 	spiderpoolv2beta1 "github.com/spidernet-io/spiderpool/pkg/k8s/apis/spiderpool.spidernet.io/v2beta1"
@@ -39,12 +39,12 @@ var _ = Describe("MacvlanOverlayOne", Label("overlay", "one-nic", "coordinator")
 			defer GinkgoRecover()
 			var annotations = make(map[string]string)
 
-			task = new(spiderdoctorV1.Nethttp)
-			plan = new(spiderdoctorV1.SchedulePlan)
-			target = new(spiderdoctorV1.NethttpTarget)
-			targetAgent = new(spiderdoctorV1.TargetAgentSepc)
-			request = new(spiderdoctorV1.NethttpRequest)
-			condition = new(spiderdoctorV1.NetSuccessCondition)
+			task = new(kdoctorV1beta1.NetReach)
+			plan = new(kdoctorV1beta1.SchedulePlan)
+			targetAgent = new(kdoctorV1beta1.NetReachTarget)
+			request = new(kdoctorV1beta1.NetHttpRequest)
+			schedule = new(kdoctorV1beta1.SchedulePlan)
+			condition = new(kdoctorV1beta1.NetSuccessCondition)
 			name = "one-macvlan-overlay-" + tools.RandomName()
 
 			annotations[common.MultusNetworks] = fmt.Sprintf("%s/%s", common.MultusNs, common.MacvlanVlan100)
@@ -56,17 +56,17 @@ var _ = Describe("MacvlanOverlayOne", Label("overlay", "one-nic", "coordinator")
 				annotations[constant.AnnoPodIPPool] = `{"interface": "net1", "ipv6": ["vlan100-v6"]}`
 			}
 
-			GinkgoWriter.Printf("update spiderdoctoragent annotation: %v/%v annotation: %v \n", common.SpiderDoctorAgentNs, common.SpiderDoctorAgentDSName, annotations)
-			spiderDoctorAgent, err = frame.GetDaemonSet(common.SpiderDoctorAgentDSName, common.SpiderDoctorAgentNs)
+			GinkgoWriter.Printf("update kdoctoragent annotation: %v/%v annotation: %v \n", common.KDoctorAgentNs, common.KDoctorAgentDSName, annotations)
+			kdoctorAgent, err = frame.GetDaemonSet(common.KDoctorAgentDSName, common.KDoctorAgentNs)
 			Expect(err).NotTo(HaveOccurred())
-			Expect(spiderDoctorAgent).NotTo(BeNil())
+			Expect(kdoctorAgent).NotTo(BeNil())
 
-			GinkgoWriter.Printf("remove old spiderdoctor %v/%v \n", common.SpiderDoctorAgentNs, common.SpiderDoctorAgentDSName)
-			err = frame.DeleteDaemonSet(common.SpiderDoctorAgentDSName, common.SpiderDoctorAgentNs)
+			GinkgoWriter.Printf("remove old kdoctor %v/%v \n", common.KDoctorAgentNs, common.KDoctorAgentDSName)
+			err = frame.DeleteDaemonSet(common.KDoctorAgentDSName, common.KDoctorAgentNs)
 			Expect(err).NotTo(HaveOccurred())
 
 			Eventually(func() bool {
-				spiderDoctorAgentPodList, err := frame.GetPodListByLabel(spiderDoctorAgent.Spec.Template.Labels)
+				spiderDoctorAgentPodList, err := frame.GetPodListByLabel(kdoctorAgent.Spec.Template.Labels)
 				if err != nil {
 					GinkgoWriter.Printf("failed to get pod list %v,error is: %v \n", err)
 					return false
@@ -78,72 +78,76 @@ var _ = Describe("MacvlanOverlayOne", Label("overlay", "one-nic", "coordinator")
 			}, common.ResourceDeleteTimeout, common.ForcedWaitingTime).Should(BeTrue())
 
 			// issue: the object has been modified; please apply your changes to the latest version and try again
-			spiderDoctorAgent.ResourceVersion = ""
-			spiderDoctorAgent.CreationTimestamp = v1.Time{}
-			spiderDoctorAgent.UID = apitypes.UID("")
-			spiderDoctorAgent.Spec.Template.Annotations = annotations
+			kdoctorAgent.ResourceVersion = ""
+			kdoctorAgent.CreationTimestamp = v1.Time{}
+			kdoctorAgent.UID = apitypes.UID("")
+			kdoctorAgent.Spec.Template.Annotations = annotations
 
-			GinkgoWriter.Printf("create spiderdoctor %v/%v \n", common.SpiderDoctorAgentNs, common.SpiderDoctorAgentDSName)
-			err = frame.CreateDaemonSet(spiderDoctorAgent)
+			GinkgoWriter.Printf("create kdoctor %v/%v \n", common.KDoctorAgentNs, common.KDoctorAgentDSName)
+			err = frame.CreateDaemonSet(kdoctorAgent)
 			Expect(err).NotTo(HaveOccurred())
 
 			nodeList, err := frame.GetNodeList()
 			Expect(err).NotTo(HaveOccurred())
 
 			ctx, cancel := context.WithTimeout(context.Background(), common.PodReStartTimeout)
 			defer cancel()
-			err = frame.WaitPodListRunning(spiderDoctorAgent.Spec.Selector.MatchLabels, len(nodeList.Items), ctx)
+			err = frame.WaitPodListRunning(kdoctorAgent.Spec.Selector.MatchLabels, len(nodeList.Items), ctx)
 			Expect(err).NotTo(HaveOccurred())
 		})
 
-		It("spiderdoctor connectivity should be succeed", Serial, Label("C00002"), Label("ebpf"), func() {
+		It("kdoctor connectivity should be succeed", Serial, Label("C00002"), Label("ebpf"), func() {
 
-			// create task spiderdoctor crd
+			enable := true
+			disable := false
+			// create task kdoctor crd
 			task.Name = name
 			// schedule
-			plan.StartAfterMinute = 0
 			plan.RoundNumber = 2
-			plan.IntervalMinute = 2
-			plan.TimeoutMinute = 2
 			task.Spec.Schedule = plan
 			// target
-			targetAgent.TestIngress = false
-			targetAgent.TestEndpoint = true
-			targetAgent.TestClusterIp = true
-			targetAgent.TestMultusInterface = frame.Info.MultusEnabled
-			targetAgent.TestNodePort = true
+			targetAgent.Ingress = &disable
+			targetAgent.Endpoint = &enable
+			targetAgent.ClusterIP = &enable
+			targetAgent.MultusInterface = &frame.Info.MultusEnabled
+			targetAgent.NodePort = &enable
 
-			targetAgent.TestIPv4 = &frame.Info.IpV4Enabled
+			targetAgent.IPv4 = &frame.Info.IpV4Enabled
 			if common.CheckCiliumFeatureOn() {
 				// TODO(tao.yang), set testIPv6 to false, reference issue: https://github.com/spidernet-io/spiderpool/issues/2007
-				testIPv6 := false
-				targetAgent.TestIPv6 = &testIPv6
+				targetAgent.IPv6 = &disable
 			} else {
-				targetAgent.TestIPv6 = &frame.Info.IpV6Enabled
+				targetAgent.IPv6 = &frame.Info.IpV6Enabled
 			}
 
-			GinkgoWriter.Printf("targetAgent for spiderdoctor %+v", targetAgent)
-			target.TargetAgent = targetAgent
-			task.Spec.Target = target
+			GinkgoWriter.Printf("targetAgent for kdoctor %+v", targetAgent)
+			task.Spec.Target = targetAgent
 
 			// request
 			request.DurationInSecond = 5
 			request.QPS = 1
 			request.PerRequestTimeoutInMS = 15000
 			task.Spec.Request = request
 
+			// Schedule
+			crontab := "0 1"
+			schedule.Schedule = &crontab
+			schedule.RoundNumber = 1
+			schedule.RoundTimeoutMinute = 1
+			task.Spec.Schedule = schedule
+
 			// success condition
 			condition.SuccessRate = &successRate
 			condition.MeanAccessDelayInMs = &delayMs
 			task.Spec.SuccessCondition = condition
 
 			taskCopy := task
-			GinkgoWriter.Printf("spiderdoctor task: %+v", task)
+			GinkgoWriter.Printf("kdoctor task: %+v", task)
 			err := frame.CreateResource(task)
-			Expect(err).NotTo(HaveOccurred(), " spiderdoctor nethttp crd create failed")
+			Expect(err).NotTo(HaveOccurred(), " kdoctor nethttp crd create failed")
 
 			err = frame.GetResource(apitypes.NamespacedName{Name: name}, taskCopy)
-			Expect(err).NotTo(HaveOccurred(), " spiderdoctor nethttp crd get failed")
+			Expect(err).NotTo(HaveOccurred(), " kdoctor nethttp crd get failed")
 
 			ctx, cancel := context.WithTimeout(context.Background(), common.KdoctorCheckTime)
 			defer cancel()
@@ -152,13 +156,13 @@ var _ = Describe("MacvlanOverlayOne", Label("overlay", "one-nic", "coordinator")
 				select {
 				case <-ctx.Done():
 					run = false
-					Expect(errors.New("wait nethttp test timeout")).NotTo(HaveOccurred(), " running spiderdoctor task timeout")
+					Expect(errors.New("wait nethttp test timeout")).NotTo(HaveOccurred(), " running kdoctor task timeout")
 				default:
 					err = frame.GetResource(apitypes.NamespacedName{Name: name}, taskCopy)
-					Expect(err).NotTo(HaveOccurred(), "spiderdoctor nethttp crd get failed,err is %v", err)
+					Expect(err).NotTo(HaveOccurred(), "kdoctor nethttp crd get failed,err is %v", err)
 
 					if taskCopy.Status.Finish == true {
-						GinkgoWriter.Printf("spiderdoctor's nethttp execution result %+v", taskCopy)
+						GinkgoWriter.Printf("kdoctor's nethttp execution result %+v", taskCopy)
 						for _, v := range taskCopy.Status.History {
 							if v.Status != "succeed" {
 								err = errors.New("error has occurred")