Upgrade k8s libraries to v0.29.2 #5843

hjiajing · 2024-01-04T14:50:13Z

Upgrade k8s libraries to v0.29.2.
Upgrade controller-runtime to v0.16.3

hjiajing · 2024-01-04T14:54:33Z

Hi @tnqn , Could you please help to view this PR. It upgrades k8s libraries from 0.26.4 to 0.29.0. Major changes are about the function wait.Poll and some other Poll functions.

And the controller-runtime is also upgraded to v0.16.3, it affects the multicluster implementation. Hi @luolanzone , Could you help review this PR, thanks.

luolanzone · 2024-01-05T02:15:05Z

/test-multicluster-e2e

luolanzone · 2024-01-05T01:49:58Z

build/images/codegen/Dockerfile

@@ -32,11 +32,11 @@ LABEL description="A Docker image based on the golang image, which includes code

 ENV GO111MODULE=on

-ARG K8S_VERSION=1.26.4
+ARG K8S_VERSION=1.29.0


Please search 1.26.4 in antrea repo, I noticed there a few other files need to be updated.
For new version of codegen image, please refer to build/images/codegen/README.md to build a new image, we need @tnqn or @antoninbas's help to upload it.

Updated codegen's README and hack/update-codegen.sh from v1.26.4 to v1.29.0.

make sure to change this to 1.29.2

and update the README

DONE. Replace v0.26.4 and v0.29.0 to 0.29.2.

luolanzone · 2024-01-05T02:22:09Z

multicluster/test/integration/suite_test.go

@@ -42,7 +44,6 @@ import (
 	antreamcscheme "antrea.io/antrea/multicluster/pkg/client/clientset/versioned/scheme"
 	antreascheme "antrea.io/antrea/pkg/client/clientset/versioned/scheme"
 	"antrea.io/antrea/pkg/signals"
-	//+kubebuilder:scaffold:imports


Why to move this line?
I think it was generated automatically before, please help to check if possible to remove it, or it has a specific purpose. Thanks.

This comment is the anchor point for kubebuilder to insert code. And I did not remove it, it was formatted and moved in front of these lines. I have moved it back.

luolanzone · 2024-01-05T02:32:12Z

pkg/agent/multicluster/stretched_networkpolicy_controller.go

@@ -184,7 +184,7 @@ func (s *StretchedNetworkPolicyController) processNextWorkItem() bool {

 	if podRef, ok := obj.(types.NamespacedName); !ok {
 		s.queue.Forget(obj)
-		klog.Errorf("Expected type 'NamespacedName' in work queue but got object", "object", obj)
+		klog.Errorf("Expected type 'NamespacedName' in work queue but got object %v\n", obj)


No need to append "\n", klog.Errorf will handle it.

use structured logging (klog.ErrorS)

third_party/proxy/service.go

pkg/agent/util/net_windows.go

pkg/agent/controller/egress/egress_controller_test.go

luolanzone

LTGM overall, two nits.

luolanzone · 2024-01-08T03:36:35Z

cmd/antrea-agent/agent.go

@@ -744,10 +744,10 @@ func run(o *Options) error {
 		// Service would fail.
 		if o.config.AntreaProxy.ProxyAll {
 			klog.InfoS("Waiting for AntreaProxy to be ready")
-			if err := wait.PollUntil(time.Second, func() (bool, error) {
+			if err := wait.PollUntilContextCancel(ctx, time.Second, true, func(ctx context.Context) (bool, error) {


Should this "immediate" be true for PollUtil?

same question
let's keep the same behavior

Yes, it should be set to false, I missed this one. Fixed.

third_party/proxy/service.go

luolanzone · 2024-01-08T03:41:23Z

/test-multicluster-e2e

antoninbas

Out of curiosity, what are the nolint:staticcheck pragmas for?

For contexts, I would follow this rule:

use context.Background() when appropriate, in particular in unit tests
use context.TODO() when we may want to add an actual context object in the future (e.g., in the future maybe the function should take a context as parameter)

In general, I don't think we should have context.Background() in a pkg function.

antoninbas · 2024-01-08T22:55:05Z

cmd/antrea-agent/agent.go

@@ -744,10 +744,10 @@ func run(o *Options) error {
 		// Service would fail.
 		if o.config.AntreaProxy.ProxyAll {
 			klog.InfoS("Waiting for AntreaProxy to be ready")
-			if err := wait.PollUntil(time.Second, func() (bool, error) {
+			if err := wait.PollUntilContextCancel(ctx, time.Second, true, func(ctx context.Context) (bool, error) {


same question
let's keep the same behavior

antoninbas · 2024-01-08T23:07:20Z

multicluster/cmd/multicluster-controller/controller.go

 	if o.SelfSignedCert {
-		o.options.CertDir = selfSignedCertDir
+		o.options.Metrics.CertDir = selfSignedCertDir
 	} else {
-		o.options.CertDir = certDir
+		o.options.Metrics.CertDir = certDir
 	}


I am not sure that's correct. The historic CertDir option seems to have been used for both the metrics server and for the webhook server. Do we need to configure WebhookServer.CertDir as well for the manager?

cc @luolanzone

Yes @antoninbas is right, the historic CertDir is for webhook according to the comment. @hjiajing please help to double check and refine this part with correct fields. Thanks.

Used webhook.NewServer to initialize the webhook server with certDir.

antoninbas · 2024-01-08T23:08:35Z

multicluster/controllers/multicluster/member/gateway_controller_test.go

@@ -256,13 +257,13 @@ func TestClusterSetMapFunc_Gateway(t *testing.T) {
 	}
 	fakeClient := fake.NewClientBuilder().WithScheme(common.TestScheme).WithObjects(clusterSet, gw1).Build()
 	r := NewGatewayReconciler(fakeClient, common.TestScheme, "default", []string{"10.200.1.1/16"}, nil)
-	requests := r.clusterSetMapFunc(clusterSet)
+	requests := r.clusterSetMapFunc(context.TODO(), clusterSet)


for unit tests, I usually suggest defining ctx := context.Background() at the beginning of the test function, then using ctx throughout the test definition

Changed, as well as other 3 files.

antoninbas · 2024-01-08T23:11:20Z

pkg/agent/agent.go

-				}
-				return true, nil
-			})
+			err = wait.PollUntilContextTimeout(context.Background(), 200*time.Millisecond, 10*time.Second, true,


maybe context.TODO() here?
same for other instances in this file

Done, changed to context.TODO().

antoninbas · 2024-01-08T23:11:57Z

pkg/agent/agent_linux.go

-		klog.InfoS("OVS bridge local port is ready", "type", link.Type(), "attrs", link.Attrs())
-		return true, nil
-	})
+	wait.PollUntilContextTimeout(context.Background(), 100*time.Millisecond, 10000*time.Millisecond, true,


antoninbas · 2024-01-08T23:12:10Z

pkg/agent/cniserver/interface_configuration_windows.go

-			currentEP, ok := ic.getEndpoint(endpointName)
-			if !ok {
-				klog.InfoS("HNSEndpoint doesn't exist in cache, exit current goroutine", "HNSEndpoint", endpointName)
+		pollErr := wait.PollUntilContextTimeout(context.Background(), 100*time.Millisecond, 60*time.Second, true,


antoninbas · 2024-01-08T23:15:00Z

pkg/agent/controller/egress/egress_controller_test.go

@@ -1118,7 +1118,7 @@ func TestSyncEgress(t *testing.T) {
 			if tt.newLocalIPs != nil {
 				c.localIPDetector = &fakeLocalIPDetector{localIPs: tt.newLocalIPs}
 			}
-			assert.NoError(t, wait.Poll(time.Millisecond*100, time.Second, func() (done bool, err error) {
+			assert.NoError(t, wait.PollUntilContextTimeout(context.Background(), time.Millisecond*100, time.Second, false, func(ctx context.Context) (done bool, err error) {


this may be a good opportunity to update these to assert.Eventually / require.Eventually

what do you think @tnqn ?

Taking into account of course this shortcoming: stretchr/testify#1396

makes sense to me

antoninbas · 2024-01-08T23:16:16Z

pkg/agent/multicluster/stretched_networkpolicy_controller.go

@@ -184,7 +184,7 @@ func (s *StretchedNetworkPolicyController) processNextWorkItem() bool {

 	if podRef, ok := obj.(types.NamespacedName); !ok {
 		s.queue.Forget(obj)
-		klog.Errorf("Expected type 'NamespacedName' in work queue but got object", "object", obj)
+		klog.Errorf("Expected type 'NamespacedName' in work queue but got object %v\n", obj)


use structured logging (klog.ErrorS)

antoninbas · 2024-01-12T20:10:21Z

Out of curiosity, what are the nolint:staticcheck pragmas for?

@hjiajing did you answer this anywhere?

antoninbas · 2024-01-12T20:11:23Z

multicluster/controllers/multicluster/member/labelidentity_controller_test.go

@@ -246,7 +247,7 @@ func TestNamespaceMapFunc(t *testing.T) {
 	mcReconciler.SetRemoteCommonArea(commonArea)

 	r := NewLabelIdentityReconciler(fakeClient, common.TestScheme, mcReconciler, "default")
-	actualReq := r.namespaceMapFunc(ns)
+	actualReq := r.namespaceMapFunc(context.TODO(), ns)


use context.Background() like in multicluster/controllers/multicluster/member/gateway_controller_test.go?

Done, changed all context.TODO() in unit tests to context.Backgrond().

I am not sure this was addressed. I still see context.TODO() in this file

antoninbas · 2024-01-12T20:13:17Z

pkg/agent/cniserver/ipam/antrea_ipam_test.go

@@ -376,7 +377,7 @@ func TestAntreaIPAMDriver(t *testing.T) {

 		podNamespace := string(k8sArgsMap[test].K8S_POD_NAMESPACE)
 		podName := string(k8sArgsMap[test].K8S_POD_NAME)
-		err = wait.Poll(time.Millisecond*200, time.Second, func() (bool, error) {
+		err = wait.PollUntilContextTimeout(context.TODO(), time.Millisecond*200, time.Second, false, func(ctx context.Context) (bool, error) {


context.Background() in unit tests

antoninbas · 2024-01-12T20:13:57Z

pkg/agent/controller/egress/egress_controller_test.go

@@ -1118,7 +1118,7 @@ func TestSyncEgress(t *testing.T) {
 			if tt.newLocalIPs != nil {
 				c.localIPDetector = &fakeLocalIPDetector{localIPs: tt.newLocalIPs}
 			}
-			assert.NoError(t, wait.Poll(time.Millisecond*100, time.Second, func() (done bool, err error) {
+			assert.NoError(t, wait.PollUntilContextTimeout(context.TODO(), time.Millisecond*100, time.Second, false, func(ctx context.Context) (done bool, err error) {


did you see my earlier comment: #5843 (comment)?

Sorry I forget to reply it, I saw it and I'm working on this part. Thanks for the reminder.

hjiajing · 2024-01-29T13:59:36Z

Out of curiosity, what are the nolint:staticcheck pragmas for?

@hjiajing did you answer this anywhere?

Because the config is deprecated in controller-runtime, but it is not used in our code, just some declarations. So I added a nolint:staticcheck for them.

antoninbas · 2024-01-30T22:01:31Z

Out of curiosity, what are the nolint:staticcheck pragmas for?

@hjiajing did you answer this anywhere?

Because the config is deprecated in controller-runtime, but it is not used in our code, just some declarations. So I added a nolint:staticcheck for them.

I see the following recommendation in the package documentation:

Deprecated: The component config package has been deprecated and will be removed in a future release. Users should migrate to their own config implementation, please share feedback in kubernetes-sigs/controller-runtime#895.

Do we have a plan to migrate to our own config implementation? If it is straightforward, should we do it as part of this PR? cc @luolanzone

luolanzone · 2024-02-01T02:34:12Z

Out of curiosity, what are the nolint:staticcheck pragmas for?

@hjiajing did you answer this anywhere?

Because the config is deprecated in controller-runtime, but it is not used in our code, just some declarations. So I added a nolint:staticcheck for them.

I see the following recommendation in the package documentation:

Deprecated: The component config package has been deprecated and will be removed in a future release. Users should migrate to their own config implementation, please share feedback in kubernetes-sigs/controller-runtime#895.

Do we have a plan to migrate to our own config implementation? If it is straightforward, should we do it as part of this PR? cc @luolanzone

Hi @antoninbas, we have one spec ControllerManagerConfigurationSpec as an inline field in MultiClusterConfig, but I think it is not used in our MC scenarios, we probably can remove config.ControllerManagerConfigurationSpec directly. @hjiajing could you double check if it's Ok to remove directly? Thanks.
Btw, please check and resolve code conflicts.

pkg/agent/agent.go

hjiajing · 2024-02-25T08:33:21Z

Hi @antoninbas , I have changed assert.NoError(t, wait.xxx) to assert.Eventually, could you please take a look, thanks.

antoninbas

Sorry for the delay.
I think we should update the PR to use 0.29.2 now, even if that means pushing a new container image for codegen. There is also a merge conflict.
Let's try to get this PR merged soon

antoninbas · 2024-02-27T23:48:38Z

multicluster/cmd/multicluster-controller/options.go

+	if *multiclusterConfig.LeaderElection.LeaderElect {
+		o.options.LeaderElection = true
+	}


I just took a quick look and it seems there are as lot of fields in multiclusterConfig.LeaderElection besides LeaderElection. I guess we are just ignoring these fields? Do we even need / want to support leader election here? I assume it's not something we leverage in the installation manifests.

Should we just include the fields we actually support in MultiClusterConfig?
Alternatively, is there a standard way of converting from configv1alpha1.LeaderElectionConfiguration to manager options? I took a look at the following PR, and it seems that they are copying fields one by one:
https://github.com/kubernetes-sigs/kueue/pull/889/files

cc @luolanzone

We actually didn't use LeaderElection in MC. I think it's stale codes which can be removed. @hjiajing could you help to remove it? Please also remove the leaderElection in the manifests, Thanks.

antoninbas · 2024-02-27T23:55:39Z

multicluster/controllers/multicluster/member/labelidentity_controller_test.go

@@ -246,7 +247,7 @@ func TestNamespaceMapFunc(t *testing.T) {
 	mcReconciler.SetRemoteCommonArea(commonArea)

 	r := NewLabelIdentityReconciler(fakeClient, common.TestScheme, mcReconciler, "default")
-	actualReq := r.namespaceMapFunc(ns)
+	actualReq := r.namespaceMapFunc(context.TODO(), ns)


I am not sure this was addressed. I still see context.TODO() in this file

antoninbas · 2024-02-27T23:58:56Z

pkg/agent/controller/egress/egress_controller_test.go

-	require.NoError(t, wait.PollImmediate(10*time.Millisecond, time.Second, func() (done bool, err error) {
-		return c.queue.Len() == 1, nil
-	}))
+	require.NoError(t, wait.PollUntilContextTimeout(context.Background(), 10*time.Millisecond, time.Second, true,


maybe require.Eventually then, given that you have made the change for assert?

same comment in other places as well

I am not sure this was addressed. I still see context.TODO() in this file

Sorry I missed this file. Changed to context.Background(). And all require.NoError(t, wait.XXX) were changed to require.Eventually.

antoninbas · 2024-02-27T23:59:53Z

pkg/agent/controller/networkpolicy/allocator_test.go

-			err := wait.PollImmediate(10*time.Millisecond, 100*time.Millisecond, func() (bool, error) {
-				return ruleHasBeenDeleted(), nil
-			})
+			err := wait.PollUntilContextTimeout(context.Background(), 10*time.Millisecond, 100*time.Millisecond, true,
+				func(ctx context.Context) (bool, error) {
+					return ruleHasBeenDeleted(), nil
+				})
 			require.Error(t, err, "Rule ID was unexpectedly released")


same comment: could use require.Eventually

Looks like you didn't push the code? I still see wait.PollUntilContextTimeout + require.Error

Sorry I missed this file, fixed. And I search globally, this file is the only one which uses wait.PollUntilContextTimeout + require.Error. Replaced this with require.Eventually.

antoninbas · 2024-02-28T00:00:21Z

pkg/agent/controller/networkpolicy/allocator_test.go

+			err = wait.PollUntilContextTimeout(context.Background(), 10*time.Millisecond, 1*time.Second, true,
+				func(ctx context.Context) (bool, error) {
+					return ruleHasBeenDeleted(), nil
+				})
 			require.NoError(t, err, "Rule ID was not released")


same thing here, I haven't checked the rest of the code

antoninbas · 2024-02-28T00:06:12Z

pkg/apiserver/certificate/certificate_test.go

@@ -199,7 +200,7 @@ func TestApplyServerCert(t *testing.T) {

 			if tt.selfSignedCert && tt.testRotate {
 				oldCertKeyContent := got.getCertificate()
-				err := wait.Poll(time.Second, 8*time.Second, func() (bool, error) {
+				err := wait.PollUntilContextTimeout(context.Background(), time.Second, 8*time.Second, false, func(ctx context.Context) (bool, error) {


use assert.Eventually

antoninbas · 2024-02-28T00:07:11Z

pkg/apiserver/registry/system/supportbundle/rest.go

@@ -308,6 +308,10 @@ func (r *supportBundleREST) clean(ctx context.Context, bundlePath string, durati
 	defaultFS.Remove(bundlePath)
 }

+func (r *supportBundleREST) GetSingularName() string {
+	return "supportBundle"


Unless I am mistaken, this should be "supportbundle", not "supportBundle"

antoninbas · 2024-02-28T00:27:47Z

test/integration/ovs/ofctrl_test.go

+		require.NoError(t, wait.PollUntilContextTimeout(context.Background(), time.Millisecond*100, time.Second, true,
+			func(ctx context.Context) (done bool, err error) {
+				flowList, err = OfctlDumpTableFlowsWithoutName(ovsCtlClient, myTable.GetID())
+				require.Nil(t, err)


not from this PR, but should be require.NoError

same for other tests below

Changed all require.Nil(t, err) to require.NoError(t, err)

hjiajing · 2024-03-15T11:14:37Z

@hjiajing The Validate metrics in Prometheus CI job failed twice in a row, probably worth checking locally Please push new changes in a separate commit, so they can be reviewed easily. We will squash when we merge.

OK, all feature gates enabled E2E failed, I'm triaging this.

antoninbas · 2024-03-15T17:11:19Z

@hjiajing that failure may be unrelated to your PR. However, the Validate metrics in Prometheus job failure is likely to be caused by your changes. After updating the K8s libraries, you usually need to regenerate the list of exposed metrics, IIRC.

hjiajing · 2024-03-15T17:35:04Z

@hjiajing that failure may be unrelated to your PR. However, the Validate metrics in Prometheus job failure is likely to be caused by your changes. After updating the K8s libraries, you usually need to regenerate the list of exposed metrics, IIRC.

Sure, I will regenerate this doc.

hjiajing · 2024-03-15T18:25:06Z

Doc updated.

pkg/apiserver/apiserver.go

pkg/apiserver/registry/networkpolicy/clustergroupmember/rest.go

pkg/apiserver/registry/networkpolicy/groupmember/rest.go

1. Upgrade k8s libraries to v0.29.2 2. Upgrade controller-runtime to v0.16.3 Signed-off-by: hjiajing <hjiajing@vmware.com>

tnqn

LGTM

tnqn · 2024-03-20T07:06:18Z

/test-all

Fix antrea-io#6129 In the failure tests, the following function is called to verify whether a connection should be allowed or denied. To verify a connection should be denied, it requires 5 seconds. ``` func probeClientIPFromPod(data *TestData, pod, container string, baseUrl string) (string, error) { url := fmt.Sprintf("%s/%s", baseUrl, "clientip") hostPort, _, err := data.runWgetCommandFromTestPodWithRetry(pod, data.testNamespace, container, url, 5) if err != nil { return "", err } host, _, err := net.SplitHostPort(hostPort) return host, err } ``` Before antrea-io#5843, these e2e tests utilized the function PollImmediate from k8s.io/apimachinery/pkg/util/wait, which immediately calls an anonymous function including the above function. Since the timeout is 5 seconds, and the ticker time is 1 second, and the anonymous function runs immediately, the 5-second timeout is sufficient to verify the denied state of a connection as mentioned above. However, after antrea-io#5843, the function `Eventually` from github.com/stretchr/testify/assert is used with the same parameters, which implies that the anonymous function runs after the first ticker time, leaving 4 seconds. 4 seconds are insufficient to verify the denied state of a connection. To resolve the issue, the timeout should be adjusted to be more than 5 seconds. Signed-off-by: Hongliang Liu <lhongliang@vmware.com>

Fix antrea-io#6129 In the failure tests, the following function is called to verify whether a connection should be allowed or denied. To verify a connection should be denied, it requires 5 seconds. ```go func probeClientIPFromPod(data *TestData, pod, container string, baseUrl string) (string, error) { url := fmt.Sprintf("%s/%s", baseUrl, "clientip") hostPort, _, err := data.runWgetCommandFromTestPodWithRetry(pod, data.testNamespace, container, url, 5) if err != nil { return "", err } host, _, err := net.SplitHostPort(hostPort) return host, err } ``` Before antrea-io#5843, these e2e tests utilized the function `PollImmediate` from `k8s.io/apimachinery/pkg/util/wait`, which immediately calls an anonymous function including the above function. Since the timeout is 5 seconds, and the ticker time is 1 second, and the anonymous function runs immediately, the 5-second timeout is sufficient to verify the denied state of a connection as mentioned above. However, after antrea-io#5843, the function `Eventually` from `github.com/stretchr/testify/assert` is used with the same parameters, which implies that the anonymous function runs after the first ticker time, leaving 4 seconds. 4 seconds are insufficient to verify the denied state of a connection. To resolve the issue, `RunCommandFromPod` called in `data.runWgetCommandFromTestPodWithRetry` is called directly in function `Eventually` to verify the connection state. Signed-off-by: Hongliang Liu <lhongliang@vmware.com>

After switching from `wait.PollImmediate` to `assert.Eventually` in #5843, the probe used to validate L7 NP enforcement was no longer correct. We improve the validation logic so that each iteration of the condition function in `assert.Eventually` only sends an HTTP probe once, instead of using a probe with its own retry mechanism. This fixes the issue. Fixes #6129 Signed-off-by: Hongliang Liu <lhongliang@vmware.com>

We re-apply antrea-io#5703. When updating K8s dependencies in antrea-io#5843, the change was reverted and we went back to using an older version of opentelemetry which is affected by a CVE. The CVE doesn't affect Antrea, the patch is meant to avoid warnings from CVE scanners. Co-authored-by: Antonin Bas <antonin.bas@broadcom.com> Signed-off-by: Bin Liu <biliu@vmware.com> Signed-off-by: Antonin Bas <antonin.bas@broadcom.com>

A few issues were introduced by antrea-io#5843 because of changes in the sigs.k8s.io/controller-runtime interface. The biggest issue was that the call to ctrl.NewManager was not using the Options object populated earlier in the setupManagerAndCertController function. Instead it was creating and using a new, incomplete Options object. Fixes antrea-io#6149 Signed-off-by: Antonin Bas <antonin.bas@broadcom.com>

A few issues were introduced by #5843 because of changes in the sigs.k8s.io/controller-runtime interface. The biggest issue was that the call to ctrl.NewManager was not using the Options object populated earlier in the setupManagerAndCertController function. Instead it was creating and using a new, incomplete Options object. Additionally, the decoder is no longer injected automatically, it needs to be instantiated by us. Otherwise the admission webhook panics. See kubernetes-sigs/controller-runtime#2695 Fixes #6149 Signed-off-by: Antonin Bas <antonin.bas@broadcom.com>

We re-apply #5703. When updating K8s dependencies in #5843, the change was reverted and we went back to using an older version of opentelemetry which is affected by a CVE. The CVE doesn't affect Antrea, the patch is meant to avoid warnings from CVE scanners. Signed-off-by: Bin Liu <biliu@vmware.com> Signed-off-by: Antonin Bas <antonin.bas@broadcom.com> Co-authored-by: Bin Liu <biliu@vmware.com>

hjiajing requested review from tnqn and luolanzone January 4, 2024 14:50

luolanzone reviewed Jan 5, 2024

View reviewed changes

hjiajing force-pushed the upgrade-k8s.io branch 2 times, most recently from d61dc7b to fd8e53b Compare January 5, 2024 05:36

luolanzone mentioned this pull request Jan 5, 2024

Compatibility with K8s 1.29 #5792

Closed

5 tasks

hjiajing force-pushed the upgrade-k8s.io branch 3 times, most recently from 141f6e9 to a4678e1 Compare January 6, 2024 15:03

luolanzone reviewed Jan 8, 2024

View reviewed changes

luolanzone requested a review from antoninbas January 8, 2024 03:43

antoninbas reviewed Jan 8, 2024

View reviewed changes

hjiajing force-pushed the upgrade-k8s.io branch 2 times, most recently from a34764f to 35fbc93 Compare January 12, 2024 02:35

antoninbas reviewed Jan 12, 2024

View reviewed changes

hjiajing force-pushed the upgrade-k8s.io branch from 35fbc93 to 0895fe5 Compare January 29, 2024 08:46

tnqn reviewed Feb 1, 2024

View reviewed changes

pkg/agent/agent.go Show resolved Hide resolved

hjiajing force-pushed the upgrade-k8s.io branch 2 times, most recently from f400885 to 2320478 Compare February 5, 2024 09:54

hjiajing force-pushed the upgrade-k8s.io branch 2 times, most recently from 437374c to e26509e Compare February 25, 2024 08:32

antoninbas reviewed Feb 28, 2024

View reviewed changes

hjiajing force-pushed the upgrade-k8s.io branch from d6073af to ccebda2 Compare March 15, 2024 08:13

hjiajing force-pushed the upgrade-k8s.io branch from ccebda2 to 2822cac Compare March 15, 2024 18:24

hjiajing force-pushed the upgrade-k8s.io branch 4 times, most recently from ae08465 to 8bbe7b0 Compare March 18, 2024 14:27

tnqn reviewed Mar 18, 2024

View reviewed changes

pkg/apiserver/apiserver.go Outdated Show resolved Hide resolved

pkg/apiserver/registry/networkpolicy/clustergroupmember/rest.go Outdated Show resolved Hide resolved

pkg/apiserver/registry/networkpolicy/groupmember/rest.go Outdated Show resolved Hide resolved

Upgrade k8s libraries to v0.29.2

77a1eb9

1. Upgrade k8s libraries to v0.29.2 2. Upgrade controller-runtime to v0.16.3 Signed-off-by: hjiajing <hjiajing@vmware.com>

hjiajing force-pushed the upgrade-k8s.io branch from 8bbe7b0 to 77a1eb9 Compare March 18, 2024 16:23

tnqn approved these changes Mar 20, 2024

View reviewed changes

antoninbas merged commit 2aafb2e into antrea-io:main Mar 20, 2024
48 of 55 checks passed

hongliangl mentioned this pull request Mar 22, 2024

Fix L7 NetworkPolicy e2e test failure #6138

Merged

This was referenced Mar 26, 2024

Bump up opentelemetry to avoid scanner CVE warnings (#5703) #6148

Merged

antrea-mc-controller no longer deploys successfully #6149

Closed

antoninbas mentioned this pull request Mar 26, 2024

[Multicluster] Fix cache options for controller Manager #6150

Merged

Upgrade k8s libraries to v0.29.2 #5843

Upgrade k8s libraries to v0.29.2 #5843

Conversation

hjiajing commented Jan 4, 2024 • edited

hjiajing commented Jan 4, 2024

luolanzone commented Jan 5, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

luolanzone left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

luolanzone commented Jan 8, 2024

antoninbas left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

antoninbas commented Jan 12, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hjiajing commented Jan 29, 2024

antoninbas commented Jan 30, 2024

luolanzone commented Feb 1, 2024 • edited

hjiajing commented Feb 25, 2024

antoninbas left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hjiajing Mar 4, 2024 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hjiajing commented Mar 15, 2024

antoninbas commented Mar 15, 2024

hjiajing commented Mar 15, 2024

hjiajing commented Mar 15, 2024

tnqn left a comment

Choose a reason for hiding this comment

tnqn commented Mar 20, 2024

hjiajing commented Jan 4, 2024 •

edited

luolanzone commented Feb 1, 2024 •

edited

hjiajing Mar 4, 2024 •

edited