[Windows] Fix agent initialization error by adjusting VMSwitch commands #3169

XinShuYang · 2022-01-04T14:26:37Z

Set/Get-VMSwitch could return hostname resolution error in some environments.
The root cause is the remote DNS name didn’t match hostname, then Set/Get-VMSwitch
could return error once it needed to get hostname from DNS name.

This issue can be fixed by adding hostname to configuration.
e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s

Signed-off-by: Shuyang Xin gavinx@vmware.com

codecov-commenter · 2022-01-04T14:32:39Z

Codecov Report

Merging #3169 (c8babe9) into main (089a008) will decrease coverage by 6.79%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main    #3169      +/-   ##
==========================================
- Coverage   58.36%   51.56%   -6.80%     
==========================================
  Files         295      297       +2     
  Lines       24907    25434     +527     
==========================================
- Hits        14536    13116    -1420     
- Misses       8792    10836    +2044     
+ Partials     1579     1482      -97

Flag	Coverage Δ
kind-e2e-tests	`30.58% <ø> (-14.09%)`	⬇️
unit-tests	`40.32% <ø> (-0.02%)`	⬇️

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

Impacted Files	Coverage Δ
pkg/controller/types/networkpolicy.go	`0.00% <0.00%> (-100.00%)`	⬇️
pkg/agent/controller/networkpolicy/reject.go	`0.00% <0.00%> (-87.91%)`	⬇️
pkg/apis/controlplane/v1beta2/helper.go	`33.33% <0.00%> (-66.67%)`	⬇️
...formers/externalversions/security/v1alpha1/tier.go	`0.00% <0.00%> (-64.29%)`	⬇️
...ers/externalversions/core/v1alpha2/clustergroup.go	`0.00% <0.00%> (-64.29%)`	⬇️
...s/externalversions/core/v1alpha2/externalentity.go	`0.00% <0.00%> (-64.29%)`	⬇️
...xternalversions/security/v1alpha1/networkpolicy.go	`0.00% <0.00%> (-64.29%)`	⬇️
...versions/security/v1alpha1/clusternetworkpolicy.go	`0.00% <0.00%> (-64.29%)`	⬇️
pkg/controller/networkpolicy/mutate.go	`0.00% <0.00%> (-62.07%)`	⬇️
pkg/agent/controller/networkpolicy/packetin.go	`6.56% <0.00%> (-60.59%)`	⬇️
... and 97 more

wenyingd

Code looks good to me, and please add more comments in the commitments to explain the issue.

XinShuYang · 2022-01-10T01:51:58Z

Code looks good to me, and please add more comments in the commitments to explain the issue.

Updated.

Set/Get-VMSwitch could return hostname resolution error in some environments. The root cause is the remote DNS name didn’t match hostname, then Set/Get-VMSwitch could return error once it needed to get hostname from DNS name. This issue can be fixed by adding hostname to configuration. e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s Signed-off-by: Shuyang Xin <gavinx@vmware.com>

wenyingd

LGTM

XinShuYang · 2022-01-13T02:03:24Z

/test-e2e

tnqn · 2022-01-14T08:14:33Z

pkg/agent/util/net_windows.go

@@ -177,7 +177,7 @@ func RemoveManagementInterface(networkName string) error {
 	var err error
 	var maxRetry = 3
 	var i = 0
-	cmd := fmt.Sprintf("Get-VMSwitch -Name %s  | Set-VMSwitch -AllowManagementOS $false ", networkName)
+	cmd := fmt.Sprintf("Get-VMSwitch -ComputerName $(hostname) -Name %s  | Set-VMSwitch -ComputerName $(hostname) -AllowManagementOS $false ", networkName)


Sorry, could you explain more about the issue? Doesn't ComputerName already default to local computer?

-ComputerName
Specifies one or more Hyper-V hosts on which the virtual switch is to be configured. NetBIOS names, IP addresses, and fully qualified domain names are allowable. The default is the local computer. Use localhost or a dot (.) to specify the local computer explicitly.

This command could fail in TKG cluster environment, and we found the different DNS name can cause this problem. Since we don't know much about Hyper-V‘s implementation，we guess it has a cache-like design to get hostname from DNS name periodically if the default name is invalid. On the contrary, using command with hostname never failed in our experiment. @tnqn

tnqn

LGTM

tnqn · 2022-01-14T14:53:20Z

/test-windows-all
/skip-networkpolicy
/skip-conformance
/skip-e2e
/skip-integration

XinShuYang · 2022-01-17T02:15:22Z

/test-windows-all

XinShuYang · 2022-01-17T14:49:56Z

/test-windows-networkpolicy

XinShuYang · 2022-01-17T14:56:07Z

/test-windows-networkpolicy

XinShuYang · 2022-01-17T14:56:35Z

/test-windows-networkpolicy

tnqn · 2022-01-17T17:31:04Z

/skip-e2e as Linux is not affected

tnqn · 2022-01-25T04:58:19Z

@XinShuYang if the issue could happen in release-1.2~1.4, could you backport it, following https://github.com/antrea-io/antrea/blob/main/docs/contributors/cherry-picks.md?

XinShuYang · 2022-01-25T06:00:42Z

Sure.

…ds (antrea-io#3169) Set/Get-VMSwitch could return hostname resolution error in some environments. The root cause is the remote DNS name didn’t match hostname, then Set/Get-VMSwitch could return error once it needed to get hostname from DNS name. This issue can be fixed by adding hostname to configuration. e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s Signed-off-by: Shuyang Xin <gavinx@vmware.com>

CI testbed building could still failed randomly with unmodified commands. This issue can be fixed by adding hostname to configuration. e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s Refer to antrea-io#3169 Signed-off-by: Shuyang Xin <gavinx@vmware.com>

CI testbed building still failed randomly with unmodified commands. This issue can be fixed by adding hostname to configuration. e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s Refer to antrea-io#3169 Signed-off-by: Shuyang Xin <gavinx@vmware.com>

CI testbed building still failed randomly with unmodified commands. This issue can be fixed by adding hostname to configuration. e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s Refer to #3169 Signed-off-by: Shuyang Xin <gavinx@vmware.com>

CI testbed building still failed randomly with unmodified commands. This issue can be fixed by adding hostname to configuration. e.g. Get-VMSwitch -ComputerName $(hostname) -Name %s Refer to antrea-io#3169 Signed-off-by: Shuyang Xin <gavinx@vmware.com>

wenyingd reviewed Jan 5, 2022

View reviewed changes

XinShuYang force-pushed the agenthns branch from c45efd9 to f2549ac Compare January 9, 2022 11:30

XinShuYang force-pushed the agenthns branch from f2549ac to c8babe9 Compare January 11, 2022 09:05

wenyingd approved these changes Jan 11, 2022

View reviewed changes

wenyingd requested a review from tnqn January 11, 2022 09:30

tnqn reviewed Jan 14, 2022

View reviewed changes

tnqn approved these changes Jan 14, 2022

View reviewed changes

tnqn merged commit c2f3670 into antrea-io:main Jan 17, 2022

tnqn added the action/backport Indicates a PR that requires backports. label Jan 25, 2022

XinShuYang mentioned this pull request Feb 23, 2022

Fix remaining VMSwitch command issue #3343

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Windows] Fix agent initialization error by adjusting VMSwitch commands #3169

[Windows] Fix agent initialization error by adjusting VMSwitch commands #3169

XinShuYang commented Jan 4, 2022 •

edited

codecov-commenter commented Jan 4, 2022 •

edited

wenyingd left a comment

XinShuYang commented Jan 10, 2022

wenyingd left a comment

XinShuYang commented Jan 13, 2022

tnqn Jan 14, 2022

XinShuYang Jan 14, 2022

tnqn left a comment

tnqn commented Jan 14, 2022

XinShuYang commented Jan 17, 2022

XinShuYang commented Jan 17, 2022

XinShuYang commented Jan 17, 2022

XinShuYang commented Jan 17, 2022

tnqn commented Jan 17, 2022

tnqn commented Jan 25, 2022

XinShuYang commented Jan 25, 2022

[Windows] Fix agent initialization error by adjusting VMSwitch commands #3169

[Windows] Fix agent initialization error by adjusting VMSwitch commands #3169

Conversation

XinShuYang commented Jan 4, 2022 • edited

codecov-commenter commented Jan 4, 2022 • edited

Codecov Report

wenyingd left a comment

Choose a reason for hiding this comment

XinShuYang commented Jan 10, 2022

wenyingd left a comment

Choose a reason for hiding this comment

XinShuYang commented Jan 13, 2022

tnqn Jan 14, 2022

Choose a reason for hiding this comment

XinShuYang Jan 14, 2022

Choose a reason for hiding this comment

tnqn left a comment

Choose a reason for hiding this comment

tnqn commented Jan 14, 2022

XinShuYang commented Jan 17, 2022

XinShuYang commented Jan 17, 2022

XinShuYang commented Jan 17, 2022

XinShuYang commented Jan 17, 2022

tnqn commented Jan 17, 2022

tnqn commented Jan 25, 2022

XinShuYang commented Jan 25, 2022

XinShuYang commented Jan 4, 2022 •

edited

codecov-commenter commented Jan 4, 2022 •

edited