Release candidate v1.4.44.4 (#2094)

* fix: assume invalid semver CNI has the required dump state command (#2078)

* fix: Updating the vmsize for e2e cilium to avoid resource scarcity (#2014)

CI: Testing the e2e test for cilium

---------

Co-authored-by: Vipul Singh <[email protected]>

.pipelines/singletenancy/cilium/cilium-e2e-step-template.yaml

@@ -32,7 +32,7 @@ steps:
         mkdir -p ~/.kube/
         echo "Create AKS cluster"
         make -C ./hack/swift azcfg AZCLI=az REGION=$(REGION_AKS_CLUSTER_TEST)
-        make -C ./hack/swift byocni-up AZCLI=az REGION=$(REGION_AKS_CLUSTER_TEST) SUB=$(SUB_AZURE_NETWORK_AGENT_TEST) CLUSTER=${{ parameters.clusterName }}-$(make revision)
+        make -C ./hack/swift byocni-up AZCLI=az REGION=$(REGION_AKS_CLUSTER_TEST) SUB=$(SUB_AZURE_NETWORK_AGENT_TEST) CLUSTER=${{ parameters.clusterName }}-$(make revision) VM_SIZE=Standard_B2ms
         echo "Cluster successfully created"
     displayName: Create test cluster
     condition: succeeded()
@@ -92,6 +92,9 @@ steps:
     displayName: "Run Azilium E2E"
 
   - script: |
+      echo "Status of the nodes and pods after the test"
+      kubectl get nodes -o wide
+      kubectl get pods -A -o wide
       echo "Logs will be available as a build artifact"
       ARTIFACT_DIR=$(Build.ArtifactStagingDirectory)/test-output/
       echo $ARTIFACT_DIR

.pipelines/singletenancy/overlay/overlay-e2e-step-template.yaml

@@ -32,7 +32,7 @@ steps:
         mkdir -p ~/.kube/
         echo "Create AKS Overlay cluster"
         make -C ./hack/swift azcfg AZCLI=az REGION=$(REGION_OVERLAY_CLUSTER_TEST)
-        make -C ./hack/swift overlay-byocni-up AZCLI=az REGION=$(REGION_OVERLAY_CLUSTER_TEST) SUB=$(SUB_AZURE_NETWORK_AGENT_TEST) CLUSTER=${{ parameters.clusterName }}-$(make revision)
+        make -C ./hack/swift overlay-byocni-up AZCLI=az REGION=$(REGION_OVERLAY_CLUSTER_TEST) SUB=$(SUB_AZURE_NETWORK_AGENT_TEST) CLUSTER=${{ parameters.clusterName }}-$(make revision) VM_SIZE=Standard_B2ms
         echo "Cluster successfully created"
     displayName: Create Overlay cluster
     condition: succeeded()
@@ -98,6 +98,9 @@ steps:
     displayName: "Run Azilium E2E on AKS Overlay"
 
   - script: |
+      echo "Status of the nodes and pods after the test"
+      kubectl get nodes -o wide
+      kubectl get pods -A -o wide
       echo "Logs will be available as a build artifact"
       ARTIFACT_DIR=$(Build.ArtifactStagingDirectory)/test-output/
       echo $ARTIFACT_DIR

cni/client/client.go

@@ -11,13 +11,16 @@ import (
 	"github.com/Azure/azure-container-networking/log"
 	"github.com/Azure/azure-container-networking/platform"
 	semver "github.com/hashicorp/go-version"
+	"github.com/pkg/errors"
 	utilexec "k8s.io/utils/exec"
 )
 
 type client struct {
 	exec utilexec.Interface
 }
 
+var ErrSemVerParse = errors.New("error parsing version")
+
 func New(exec utilexec.Interface) *client {
 	return &client{exec: exec}
 }
@@ -58,5 +61,10 @@ func (c *client) GetVersion() (*semver.Version, error) {
 		return nil, fmt.Errorf("Unexpected Azure CNI Version formatting: %v", output)
 	}
 
-	return semver.NewVersion(res[3])
+	version, versionErr := semver.NewVersion(res[3])
+	if versionErr != nil {
+		return nil, errors.Wrap(ErrSemVerParse, versionErr.Error())
+	}
+
+	return version, nil
 }

cns/cnireconciler/version.go

@@ -5,6 +5,7 @@ import (
 
 	"github.com/Azure/azure-container-networking/cni/client"
 	semver "github.com/hashicorp/go-version"
+	"github.com/pkg/errors"
 	"k8s.io/utils/exec"
 )
 
@@ -15,7 +16,9 @@ const lastCNIWithoutDumpStateVer = "1.4.6"
 // IsDumpStateVer checks if the CNI executable is a version that
 // has the dump state command required to initialize CNS from CNI
 // state and returns the result of that test or an error. Will always
-// return false when there is an error.
+// return false when there is an error unless the error was caused
+// by the CNI not being a semver, in which case we'll assume we can
+// use the command.
 func IsDumpStateVer() (bool, error) {
 	return isDumpStateVer(exec.New())
 }
@@ -28,6 +31,11 @@ func isDumpStateVer(exec exec.Interface) (bool, error) {
 	cnicli := client.New(exec)
 	ver, err := cnicli.GetVersion()
 	if err != nil {
+		// If the error was that the CNI isn't a valid semver, assume we have the
+		// the dump state command
+		if errors.Is(err, client.ErrSemVerParse) {
+			return true, nil
+		}
 		return false, fmt.Errorf("failed to invoke CNI client.GetVersion(): %w", err)
 	}
 	return ver.GreaterThan(needVer), nil

cns/cnireconciler/version_test.go

@@ -48,6 +48,18 @@ func TestIsDumpStateVer(t *testing.T) {
 			want:    true,
 			wantErr: false,
 		},
+		{
+			name:    "non-semver",
+			exec:    newCNIVersionFakeExec(`Azure CNI Version v1.4.35_Win2019OverlayFix`),
+			want:    true,
+			wantErr: false,
+		},
+		{
+			name:    "non-semver hotfix ver",
+			exec:    newCNIVersionFakeExec(`Azure CNI Version v1.4.44.4`),
+			want:    true,
+			wantErr: false,
+		},
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {

test/integration/utils_test.go

@@ -5,15 +5,16 @@ package k8s
 import (
 	"bytes"
 	"context"
-	"errors"
 	"io"
 	"log"
+	"os"
 	"strings"
+	"testing"
 	"time"
 
+	"github.com/pkg/errors"
+
 	// crd "dnc/requestcontroller/kubernetes"
-	"os"
-	"testing"
 
 	"github.com/Azure/azure-container-networking/test/integration/retry"
 	apiv1 "k8s.io/api/core/v1"
@@ -215,7 +216,7 @@ func waitForPodsRunning(ctx context.Context, clientset *kubernetes.Clientset, na
 
 		for _, pod := range podList.Items {
 			if pod.Status.PodIP == "" {
-				return errors.New("a pod has not been allocated an IP")
+				return errors.Wrapf(err, "Pod %s/%s has not been allocated an IP yet with reason %s", pod.Namespace, pod.Name, pod.Status.Message)
 			}
 		}