Fix agent nil panic in pure IPv6 cluster #2655

wenqiq · 2021-08-26T17:46:43Z

Fixed agent panic because of nil NodeIP in pure IPv6 cluster

The nodeConfig.NodeIPv4Addr is nil which would cause panic in agent,
when starting agent with Egress feature enabled in pure IPv6 cluster.
Related #2436

Add Egress IPv6 test cases in dual-stack or pure IPv6 cluster. Related #2196

Signed-off-by: Wenqi Qiu wenqiq@vmware.com

wenqiq · 2021-08-26T18:00:10Z

/test-e2e /test-ipv6-only-conformance /test-ipv6-only-e2e

codecov-commenter · 2021-08-26T19:20:22Z

Codecov Report

Merging #2655 (b0a9295) into main (bfa1bc4) will decrease coverage by 0.17%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #2655      +/-   ##
==========================================
- Coverage   60.66%   60.49%   -0.18%     
==========================================
  Files         285      285              
  Lines       23006    23017      +11     
==========================================
- Hits        13957    13923      -34     
- Misses       7550     7598      +48     
+ Partials     1499     1496       -3

Flag	Coverage Δ
kind-e2e-tests	`48.20% <0.00%> (-0.17%)`	⬇️
unit-tests	`41.01% <100.00%> (-0.04%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
pkg/agent/memberlist/cluster.go	`71.63% <100.00%> (+0.07%)`	⬆️
pkg/agent/controller/traceflow/packetin.go	`59.49% <0.00%> (-5.49%)`	⬇️
...agent/controller/traceflow/traceflow_controller.go	`70.21% <0.00%> (-3.20%)`	⬇️
pkg/controller/traceflow/controller.go	`70.30% <0.00%> (-2.43%)`	⬇️
pkg/agent/cniserver/ipam/ipam_delegator.go	`48.83% <0.00%> (-2.33%)`	⬇️
pkg/agent/wireguard/client_linux.go	`65.17% <0.00%> (-2.17%)`	⬇️
pkg/controller/egress/controller.go	`84.12% <0.00%> (-1.02%)`	⬇️
pkg/agent/openflow/pipeline.go	`74.14% <0.00%> (-1.02%)`	⬇️
...g/agent/cniserver/interface_configuration_linux.go	`16.15% <0.00%> (-0.69%)`	⬇️
pkg/agent/agent.go	`50.91% <0.00%> (ø)`
... and 4 more

wenqiq · 2021-08-27T07:35:36Z

/test-e2e

/test-ipv6-only-e2e

wenqiq · 2021-08-27T18:46:42Z

/test-ipv6-only-e2e

wenqiq · 2021-08-27T19:18:22Z

/test-ipv6-only-e2e

wenqiq · 2021-08-30T08:41:13Z

/test-ipv6-only-e2e

wenqiq · 2021-08-30T10:27:26Z

/test-e2e
/test-ipv6-only-conformance
/test-ipv6-only-e2e

antoninbas

looks like this will need to be backported to release-1.3; @tnqn @wenqiq

cmd/antrea-agent/agent.go

antoninbas · 2021-08-31T02:38:25Z

test/e2e/egress_test.go

+					// skipIfNotIPv6Cluster(t)
+					if clusterInfo.podV6NetworkCIDR == "" {


isn't that the same check as skipIfNotIPv6Cluster?

Yes, good catch, will fix it.

antoninbas · 2021-08-31T02:40:19Z

test/e2e/egress_test.go

+		for _, ipR := range ipRanges {
+			ipVsersion := fmt.Sprintf("-IPv%d", ipR.ipVersion)
+			expectedTotal := ipR.expectedTotalIpNum
+			t.Run(tt.name+ipVsersion, func(t *testing.T) {


aren't you going to have several tests with the same name here? it seems the name is the same for all IPRanges sharing the same IP family.

looks like this will need to be backported to release-1.3; @tnqn @wenqiq

I think yes. Do you have any comments? @tnqn

@wenqiq You should reply comment that is not specific to a line in the comment of conversation page. Replying here is confusing and hard to track the original comment.

Got it. Thanks for reminding.

antoninbas · 2021-08-31T02:40:41Z

test/e2e/egress_test.go

+						return exists == false, nil
+					})
+					require.NoError(t, err, "Failed to check if IP exists on Node")
+					// assert.False(t, exists, "Found stale IP on Node")


remove if not needed

It seems that I comment this line by mistake, will fix it.

antoninbas · 2021-08-31T02:42:28Z

test/e2e/framework.go

+		var builder strings.Builder
+		pods, errGet := data.clientset.CoreV1().Pods(antreaNamespace).List(context.TODO(), metav1.ListOptions{
+			LabelSelector: "app=antrea,component=antrea-agent",
+		})
+		if errGet != nil {
+			builder.WriteString(errGet.Error())
+		}
+		for _, pod := range pods.Items {
+			code, stdout, stderr, errCmd := provider.RunCommandOnNode(controlPlaneNodeName(), fmt.Sprintf("kubectl -n %s logs %s antrea-agent", antreaNamespace, pod.Name))
+			builder.WriteString(fmt.Sprintf("RunCommandOnNode, code: %d, stdout: %s, stderr: %s, error: %v", code, stdout, stderr, errCmd))
+		}
+		return fmt.Errorf("restartAntreaAgentPods error： %v, logs: %s", err, builder.String())
+	}


this seems unrelated to the rest of the PR and it's unclear what the purpose is since there is no comment
if there is an issue with the e2e test framework, it should be addressed in a separate PR

When the agent starting failed it just describe the agent pods, I think we need to print agent starting log and makes it easier to debug.

Agree with @antoninbas. And we already collect the component logs in specific directory if the test fails. I'm not sure if it will collect logs when deploying fails. But even if it doesn't, it's easy to change its behavior to do it. Dumpping the whole log in the error may mess up the test output.

If we don't have the same test environment. It`s hard to find the panic cause, because the describe pod logs in Jenkins CI output just say 'Back off....'.

Normal Created 116s (x4 over 2m53s) kubelet Created container antrea-agent Normal Started 116s (x4 over 2m53s) kubelet Started container antrea-agent Warning BackOff 112s (x7 over 2m48s) kubelet Back-off restarting failed container

lzhecheng · 2021-08-31T03:40:07Z

cmd/antrea-agent/agent.go

+	} else if nodeConfig.NodeIPv6Addr != nil {
+		nodeIP = nodeConfig.NodeIPv6Addr.IP
+	} else {
+		return fmt.Errorf("NodeIPAddr in Node config invalid: %v", nodeConfig)


NodeIPAddr in Node config is invalid or invalid NodeIPAddr in Node config

pkg/agent/memberlist/cluster.go

tnqn · 2021-08-31T07:14:38Z

test/e2e/egress_test.go

-				}
-				if egress.Spec.EgressIP != tt.expectedEgressIP {
-					return false, nil
+		for _, ipR := range ipRanges {


The nested subtests don't seem necessary and hard to understand what the test focus on and expect.
Can you just add subtests for several IPv6 cases here? e.g. "single matching Node with IPv6 range". I don't think we need to cover all scenario with all address families with all possible ip range type. Just typical is fine.

tnqn · 2021-08-31T07:19:54Z

test/e2e/egress_test.go

-			require.NoError(t, err, "Failed to check if IP exists on Node")
-			assert.False(t, exists, "Found stale IP on Node")
-		})
+		for _, ipR := range ipRanges {


tnqn · 2021-08-31T07:27:15Z

test/e2e/framework.go

+		var builder strings.Builder
+		pods, errGet := data.clientset.CoreV1().Pods(antreaNamespace).List(context.TODO(), metav1.ListOptions{
+			LabelSelector: "app=antrea,component=antrea-agent",
+		})
+		if errGet != nil {
+			builder.WriteString(errGet.Error())
+		}
+		for _, pod := range pods.Items {
+			code, stdout, stderr, errCmd := provider.RunCommandOnNode(controlPlaneNodeName(), fmt.Sprintf("kubectl -n %s logs %s antrea-agent", antreaNamespace, pod.Name))
+			builder.WriteString(fmt.Sprintf("RunCommandOnNode, code: %d, stdout: %s, stderr: %s, error: %v", code, stdout, stderr, errCmd))
+		}
+		return fmt.Errorf("restartAntreaAgentPods error： %v, logs: %s", err, builder.String())
+	}


Agree with @antoninbas. And we already collect the component logs in specific directory if the test fails. I'm not sure if it will collect logs when deploying fails. But even if it doesn't, it's easy to change its behavior to do it. Dumpping the whole log in the error may mess up the test output.

wenqiq · 2021-09-01T17:10:30Z

I would suggest the following:

Fix agent panic because of nil NodeIP in pure IPv6 cluster

The nodeConfig.NodeIPv4Addr is nil which would cause panic in agent,
when starting agent with Egress feature enabled in pure IPv6 cluster.

It also adds Egress IPv6 test cases in dual-stack and pure IPv6 cluster. 

Related #2196

Done

/test-all
/test-e2e
/test-ipv6-only-conformance
/test-ipv6-only-e2e
/test-ipv6-e2e

tnqn

LGTM

wenqiq · 2021-09-01T18:44:27Z

/test-e2e

wenqiq · 2021-09-02T02:08:40Z

@tnqn @antoninbas It seems all the test checks related have finished successfully. Do you have any more comments about this PR?

antoninbas · 2021-09-02T02:54:34Z

@wenqiq I see failure for dual-stack e2e tests. Is that expected?

--- FAIL: TestEgress (120.95s)
    --- FAIL: TestEgress/testEgressClientIP (46.04s)
        --- FAIL: TestEgress/testEgressClientIP/ipv4-cluster (21.02s)
        --- PASS: TestEgress/testEgressClientIP/ipv6-cluster (25.02s)
    --- FAIL: TestEgress/testEgressCRUD (4.56s)
        --- FAIL: TestEgress/testEgressCRUD/single_matching_Node (0.75s)
        --- FAIL: TestEgress/testEgressCRUD/single_matching_Node_with_IPv6_range (0.75s)
        --- FAIL: TestEgress/testEgressCRUD/two_matching_Nodes (1.45s)
        --- PASS: TestEgress/testEgressCRUD/no_matching_Node (1.60s)
    --- PASS: TestEgress/testEgressUpdateEgressIP (5.40s)
        --- PASS: TestEgress/testEgressUpdateEgressIP/same_Node (1.80s)
        --- PASS: TestEgress/testEgressUpdateEgressIP/different_Nodes (1.80s)
        --- PASS: TestEgress/testEgressUpdateEgressIP/different_Nodes_in_IPv6_cluster (1.80s)
    --- PASS: TestEgress/testEgressUpdateNodeSelector (5.20s)
        --- PASS: TestEgress/testEgressUpdateNodeSelector/IPv4_cluster (2.60s)
        --- PASS: TestEgress/testEgressUpdateNodeSelector/IPv6_cluster (2.60s)
    --- FAIL: TestEgress/testEgressNodeFailure (3.89s)
        --- FAIL: TestEgress/testEgressNodeFailure/IPv4_cluster (1.40s)
        --- PASS: TestEgress/testEgressNodeFailure/IPv6_cluster (2.49s)

wenqiq · 2021-09-02T03:29:19Z

/test-ipv6-e2e
Retrigger. It seems dual-stack e2e tests failure related to test machine. I see some error logs as following:

=== RUN   TestEgress/testEgressNodeFailure/IPv4_cluster
    egress_test.go:554: Error when running command 'pkill -STOP antrea-agent' on Node 'antrea-ipv6-5-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-5-1' in SSH config
    egress_test.go:554: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-5-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-5-1' in SSH config
    egress_test.go:554: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-5-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-5-1' in SSH config
=== RUN   TestEgress/testEgressNodeFailure/IPv6_cluster

wenqiq · 2021-09-02T05:07:18Z

/test-ipv6-e2e

wenqiq · 2021-09-02T07:32:26Z

/test-all
/test-e2e
/test-ipv6-only-conformance
/test-ipv6-only-e2e
/test-ipv6-e2e

wenqiq · 2021-09-02T09:48:15Z

Dual-stack e2e tests failed.
/test-ipv6-e2e

=== RUN   TestEgress/testEgressNodeFailure/IPv4_cluster
    egress_test.go:575: Error when running command 'pkill -STOP antrea-agent' on Node 'antrea-ipv6-9-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-9-1' in SSH config
    egress_test.go:575: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-9-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-9-1' in SSH config
    egress_test.go:575: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-9-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-9-1' in SSH config
=== RUN   TestEgress/testEgressNodeFailure/IPv6_cluster
    egress_test.go:575: Error when running command 'pkill -STOP antrea-agent' on Node 'antrea-ipv6-9-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-9-1' in SSH config
    egress_test.go:575: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-9-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-9-1' in SSH config
    egress_test.go:575: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-9-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-9-1' in SSH config
=== CONT  TestEgress
    fixtures.go:257: Exporting test logs to '/var/lib/jenkins/workspace/antrea-ipv6-ds-e2e-for-pull-request/antrea-test-logs/TestEgress/beforeTeardown.Sep02-08-59-31'
    fixtures.go:363: Error when exporting kubelet logs: error when running journalctl on Node 'antrea-ipv6-9-0', is it available? Error: <nil>
    fixtures.go:384: Deleting 'antrea-test' K8s Namespace

wenqiq · 2021-09-02T10:41:23Z

/test-all
/test-e2e
/test-ipv6-only-conformance
/test-ipv6-only-e2e

wenqiq · 2021-09-02T12:55:04Z

/test-ipv6-only-e2e
ipv6 only e2e test failed:

=== RUN   TestEgress/testEgressNodeFailure/IPv6_cluster
    egress_test.go:575: Error when running command 'pkill -STOP antrea-agent' on Node 'antrea-ipv6-8-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-8-1' in SSH config
    egress_test.go:575: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-8-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-8-1' in SSH config
    egress_test.go:575: Error when running command 'pkill -CONT antrea-agent' on Node 'antrea-ipv6-8-1', rc: 0, stdout: , stderr: , error: unable to find 'HostName' for 'antrea-ipv6-8-1' in SSH config

wenqiq · 2021-09-02T13:48:58Z

/test-ipv6-conformance
/test-ipv6-networkpolicy
/test-ipv6-only-networkpolicy

lzhecheng · 2021-09-02T14:00:05Z

@wenqiq the failure about unable to find Hostname results from ssh-config not properly configured. In this PR, you are using runCommandFromPod to run command from a node other than master/control plane, right? Before, there's no such case in IPv6-only and dual-stack tests so there's no error. @xliuxu has a PR to support this in this PR: #2675

@tnqn Branch 1.3 seems waiting for this PR so I suggest merging Xu's PR now, ignoring the comment to add a comment in script.

tnqn · 2021-09-02T14:00:15Z

@wenqiq DCO is missing in the second patch

wenqiq · 2021-09-02T14:13:59Z

Thanks @lzhecheng @tnqn @xliuxu , It seems I should rebase and merge my two commit after #2675 merged.

wenqiq · 2021-09-02T15:25:40Z

/test-ipv6-e2e

tnqn · 2021-09-02T15:31:30Z

#2675 has been merged, you could rebase now.

The nodeConfig.NodeIPv4Addr is nil which would cause panic in agent, when starting agent with Egress feature enabled in pure IPv6 cluster. It also adds Egress IPv6 test cases in dual-stack and pure IPv6 cluster. Related antrea-io#2196 Signed-off-by: Wenqi Qiu <wenqiq@vmware.com>

tnqn · 2021-09-02T15:42:50Z

test/e2e/egress_test.go

@@ -552,7 +717,7 @@ func (data *TestData) createEgress(t *testing.T, generateName string, matchExpre
 }

 func (data *TestData) waitForEgressRealized(egress *v1alpha2.Egress) (*v1alpha2.Egress, error) {
-	err := wait.PollImmediate(200*time.Millisecond, 3*time.Second, func() (done bool, err error) {
+	err := wait.PollImmediate(200*time.Millisecond, 5*time.Second, func() (done bool, err error) {


So 3 seconds is enough in some cases? I feel it may indicate there is some problem but I don't think it's related to this PR so I'm fine with the change. We'd better to look at the logs to understand why the delay is so long.

3 seconds is enough in most cases. I have tested hundreds of times in my local env (dual-stack cluster) and nerver failed.It failed once in the jenkins CI workflow:

=== RUN TestEgress/testEgressNodeFailure/IPv6_cluster egress_test.go:611: Error Trace: egress_test.go:611 egress_test.go:589 Error: Received unexpected error: timed out waiting for the condition Test: TestEgress/testEgressNodeFailure/IPv6_cluster

So I changed it to 5 seconds.

wenqiq · 2021-09-02T16:21:56Z

/test-all
/test-e2e
/test-ipv6-only-conformance
/test-ipv6-only-e2e
/test-ipv6-e2e

/test-ipv6-conformance
/test-ipv6-networkpolicy
/test-ipv6-only-networkpolicy

tnqn · 2021-09-03T00:06:29Z

The only failure in "jenkins-ipv6-only-e2e" is not related to this PR and is going to be fixed by #2712:

=== RUN   TestWireGuard/testServiceConnectivity
    wireguard_test.go:149: 
        	Error Trace:	wireguard_test.go:149
        	            				wireguard_test.go:73
        	Error:      	Received unexpected error:
        	            	nc stdout: <>, stderr: <>, err: <command terminated with exit code 1>
        	Test:       	TestWireGuard/testServiceConnectivity
        	Messages:   	Pod hostnetwork-pod should be able to connect the service's NodePort 127.0.0.1:%!s(int32=31776), but was not able to connect
=== CONT  TestWireGuard
    fixtures.go:257: Exporting test logs to '/var/lib/jenkins/workspace/antrea-ipv6-only-e2e-for-pull-request/antrea-test-logs/TestWireGuard/beforeTeardown.Sep02-18-16-05'
    fixtures.go:363: Error when exporting kubelet logs: error when running journalctl on Node 'antrea-ipv6-8-0', is it available? Error: <nil>
    fixtures.go:384: Deleting 'antrea-test' K8s Namespace
--- FAIL: TestWireGuard (102.19s)
    --- PASS: TestWireGuard/testPodConnectivity (16.26s)
    --- FAIL: TestWireGuard/testServiceConnectivity (27.28s)

wenqiq force-pushed the ipv6-e2e branch from 42227b0 to 7e8dd03 Compare August 26, 2021 17:54

wenqiq force-pushed the ipv6-e2e branch from 7e8dd03 to 65668e0 Compare August 27, 2021 05:52

wenqiq force-pushed the ipv6-e2e branch 2 times, most recently from 1577d3a to e0a6988 Compare August 27, 2021 18:43

wenqiq force-pushed the ipv6-e2e branch from e0a6988 to e486dd2 Compare August 27, 2021 19:16

wenqiq force-pushed the ipv6-e2e branch 3 times, most recently from d56d20f to b4046c7 Compare August 30, 2021 08:40

wenqiq changed the title ~~[WIP]Add Egress IPv6 test cases~~ Add Egress IPv6 test cases Aug 30, 2021

wenqiq marked this pull request as ready for review August 30, 2021 09:59

wenqiq force-pushed the ipv6-e2e branch from b4046c7 to bf355bf Compare August 30, 2021 10:23

wenqiq changed the title ~~Add Egress IPv6 test cases~~ Fix agent nil panic in pure IPv6 cluster Aug 30, 2021

wenqiq requested review from lzhecheng, tnqn and antoninbas August 31, 2021 01:46

antoninbas reviewed Aug 31, 2021

View reviewed changes

lzhecheng reviewed Aug 31, 2021

View reviewed changes

wenqiq force-pushed the ipv6-e2e branch 3 times, most recently from de06b87 to 8d9c26b Compare August 31, 2021 04:09

tnqn reviewed Aug 31, 2021

View reviewed changes

pkg/agent/memberlist/cluster.go Show resolved Hide resolved

tnqn reviewed Aug 31, 2021

View reviewed changes

wenqiq dismissed tnqn’s stale review via 341c0b6 September 1, 2021 17:08

wenqiq force-pushed the ipv6-e2e branch from a9dd443 to 341c0b6 Compare September 1, 2021 17:08

tnqn previously approved these changes Sep 1, 2021

View reviewed changes

wenqiq dismissed tnqn’s stale review via b0a9295 September 2, 2021 07:31

wenqiq force-pushed the ipv6-e2e branch from b0a9295 to 442185b Compare September 2, 2021 14:53

wenqiq force-pushed the ipv6-e2e branch from 442185b to 0a8b58f Compare September 2, 2021 15:36

tnqn reviewed Sep 2, 2021

View reviewed changes

tnqn approved these changes Sep 2, 2021

View reviewed changes

tnqn added this to the Antrea v1.3 release milestone Sep 3, 2021

tnqn merged commit 113ad78 into antrea-io:main Sep 3, 2021

tnqn mentioned this pull request Sep 3, 2021

Automated cherry pick of #2655: Fix agent panic because of nil NodeIP in pure IPv6 cluster #2713

Merged

		// skipIfNotIPv6Cluster(t)
		if clusterInfo.podV6NetworkCIDR == "" {

Fix agent nil panic in pure IPv6 cluster #2655

Fix agent nil panic in pure IPv6 cluster #2655

Conversation

wenqiq commented Aug 26, 2021 • edited

wenqiq commented Aug 26, 2021

codecov-commenter commented Aug 26, 2021 • edited

Codecov Report

wenqiq commented Aug 27, 2021 • edited

wenqiq commented Aug 27, 2021

wenqiq commented Aug 27, 2021

wenqiq commented Aug 30, 2021

wenqiq commented Aug 30, 2021

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

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

wenqiq commented Sep 1, 2021

tnqn left a comment

Choose a reason for hiding this comment

wenqiq commented Sep 1, 2021

wenqiq commented Sep 2, 2021

antoninbas commented Sep 2, 2021

wenqiq commented Sep 2, 2021

wenqiq commented Sep 2, 2021

wenqiq commented Sep 2, 2021 • edited

wenqiq commented Sep 2, 2021

wenqiq commented Sep 2, 2021

wenqiq commented Sep 2, 2021

wenqiq commented Sep 2, 2021

lzhecheng commented Sep 2, 2021

tnqn commented Sep 2, 2021

wenqiq commented Sep 2, 2021

wenqiq commented Sep 2, 2021

tnqn commented Sep 2, 2021

Choose a reason for hiding this comment

wenqiq Sep 2, 2021 • edited

Choose a reason for hiding this comment

wenqiq commented Sep 2, 2021

tnqn commented Sep 3, 2021

wenqiq commented Aug 26, 2021 •

edited

codecov-commenter commented Aug 26, 2021 •

edited

wenqiq commented Aug 27, 2021 •

edited

wenqiq commented Sep 2, 2021 •

edited

wenqiq Sep 2, 2021 •

edited