Add DHCP IP Retries in PrepareHNSNetwork #5819

XinShuYang · 2023-12-21T04:08:31Z

To address the potential race condition issue where acquiring a DHCP IP address may fail after CreateHNSNetwork,
we added a retry mechanism to wait for an available IP. If the DHCP IP cannot be acquired within six seconds,
an error will be logged.

XinShuYang · 2023-12-21T04:09:17Z

/test-windows-containerd-e2e

XinShuYang · 2023-12-21T06:20:55Z

/test-windows-containerd-e2e

XinShuYang · 2023-12-21T06:21:47Z

/test-windows-containerd-e2e

wenyingd

Can we add a e2e test to verify that vNIC migration is not supposed to modify the original properties.

pkg/agent/util/net_windows.go

XinShuYang · 2023-12-25T01:21:35Z

/test-windows-containerd-e2e

pkg/agent/util/net_windows.go

ci/jenkins/test.sh

pkg/agent/util/net_windows.go

XinShuYang · 2023-12-28T05:54:15Z

/test-windows-containerd-e2e

wenyingd

LGTM

XinShuYang · 2023-12-28T06:13:51Z

/test-windows-containerd-e2e

XinShuYang · 2023-12-28T06:24:14Z

/test-windows-containerd-e2e

XinShuYang · 2023-12-28T07:14:02Z

/test-windows-containerd-e2e

ci/jenkins/test.sh

pkg/agent/util/net_windows.go

tnqn · 2023-12-28T09:14:37Z

pkg/agent/util/net_windows.go

+			if err != nil {
+				klog.ErrorS(err, "Failed to get Ipv4 DHCP status on the network adapter", "adapter", uplinkAdapter.Name)
+			}
+			klog.Warningf("Timeout acquiring IP for the adapter, DHCP status: %t", dhcpStatus)


Use structured logging for new logs, use InfoS if this is something expected and ErrorS if not expected to happen.

Logging dhcpStatus could be confusing when it fails to get its status.

Updated the logic, now we only print dhcpStatus if we successfully retrieve its value from adapter. @tnqn

tnqn · 2023-12-28T09:16:54Z

pkg/agent/util/net_windows.go

+	// Therefore, we set the timeout limit to triple of that value, allowing a maximum wait of 6 seconds here.
+	err = wait.PollImmediate(1*time.Second, 6*time.Second, func() (bool, error) {
+		var checkErr error
+		adapter, ipFound, checkErr = adapterIPExists(nodeIPNet.IP, uplinkAdapter.HardwareAddr, ContainerVNICPrefix)


Does Windows support both DHCP and static case? If yes, these retries would add unnecessary initialization delay for static IP case.
I think it should first check if this is DHCP case, and only expects it to get IP from DHCP server when DHCP is enabled?

In my understanding, the "IP delay available issue" occurs during the creation of a new uplink interface, and this delay is consistent regardless of whether DHCP is enabled on the interface. Regarding the static IP case, we still expect to get available IP from adapter timely. Otherwise it may indicate an issue with the adapter itself. @wenyingd please correct me if I am wrong.

For static IP configurations, we still expect Windows OS can automatically migrate the IP address from pnic to vnic. We would like to use this check to ensure that Windows OS has performed expected behavior, and gives warning logs if not.

XinShuYang · 2023-12-28T10:07:05Z

/test-windows-all

XinShuYang · 2023-12-31T06:45:31Z

/test-windows-all

tnqn · 2024-01-02T07:01:25Z

pkg/agent/util/net_windows.go

+		if err == wait.ErrWaitTimeout {
+			dhcpStatus, err := InterfaceIPv4DhcpEnabled(uplinkAdapter.Name)
+			if err != nil {
+				klog.ErrorS(err, "Failed to get Ipv4 DHCP status on the network adapter", "adapter", uplinkAdapter.Name)


Suggested change

klog.ErrorS(err, "Failed to get Ipv4 DHCP status on the network adapter", "adapter", uplinkAdapter.Name)

klog.ErrorS(err, "Failed to get IPv4 DHCP status on the network adapter", "adapter", uplinkAdapter.Name)

tnqn · 2024-01-02T07:01:37Z

pkg/agent/util/net_windows.go

@@ -647,6 +670,16 @@ func HostInterfaceExists(ifaceName string) bool {
 	return true
 }

+// InterfaceIPv4DhcpEnabled returns the Ipv4 DHCP status on the specified interface.


tnqn · 2024-01-02T07:02:55Z

In the commit message: s/a warning message will be returned/an error will be logged/

tnqn · 2024-01-02T07:03:31Z

No need to rerun tests after addressing the typos

To address the potential race condition issue where acquiring a DHCP IP address may fail after CreateHNSNetwork, we added a retry mechanism to wait for an available IP. If the DHCP IP cannot be acquired within six seconds, an error will be logged. Signed-off-by: Shuyang Xin <gavinx@vmware.com>

XinShuYang · 2024-01-02T08:23:54Z

No need to rerun tests after addressing the typos

Got it, PR has been updated.

tnqn · 2024-01-02T08:25:56Z

/skip-all

XinShuYang requested a review from wenyingd December 21, 2023 04:09

XinShuYang force-pushed the windhcp branch 5 times, most recently from a765a81 to 3aafc50 Compare December 21, 2023 06:07

XinShuYang force-pushed the windhcp branch from 3aafc50 to 44ba719 Compare December 21, 2023 07:31

wenyingd reviewed Dec 21, 2023

View reviewed changes

pkg/agent/util/net_windows.go Outdated Show resolved Hide resolved

pkg/agent/util/net_windows.go Outdated Show resolved Hide resolved

pkg/agent/util/net_windows.go Outdated Show resolved Hide resolved

XinShuYang force-pushed the windhcp branch 5 times, most recently from 7c919ee to 801139a Compare December 25, 2023 01:20

XinShuYang force-pushed the windhcp branch 3 times, most recently from 4b57901 to e912ea2 Compare December 25, 2023 02:48

wenyingd reviewed Dec 25, 2023

View reviewed changes

XinShuYang force-pushed the windhcp branch 6 times, most recently from dcccd93 to 5973c2f Compare December 26, 2023 06:57

wenyingd reviewed Dec 26, 2023

View reviewed changes

ci/jenkins/test.sh Outdated Show resolved Hide resolved

ci/jenkins/test.sh Outdated Show resolved Hide resolved

pkg/agent/util/net_windows.go Outdated Show resolved Hide resolved

pkg/agent/util/net_windows.go Outdated Show resolved Hide resolved

XinShuYang force-pushed the windhcp branch 2 times, most recently from f2b4ca2 to 4bf6142 Compare December 26, 2023 22:03

XinShuYang force-pushed the windhcp branch from 7616c33 to d15f6d8 Compare December 28, 2023 06:01

wenyingd previously approved these changes Dec 28, 2023

View reviewed changes

XinShuYang requested a review from tnqn December 28, 2023 06:13

XinShuYang dismissed wenyingd’s stale review via 0c90c24 December 28, 2023 06:48

XinShuYang force-pushed the windhcp branch from d15f6d8 to 0c90c24 Compare December 28, 2023 06:48

tnqn reviewed Dec 28, 2023

View reviewed changes

XinShuYang force-pushed the windhcp branch 3 times, most recently from f2f6c7a to 7d125ac Compare December 28, 2023 10:06

XinShuYang force-pushed the windhcp branch 2 times, most recently from a737bab to ee8222f Compare December 31, 2023 06:44

XinShuYang requested a review from tnqn January 2, 2024 05:42

tnqn previously approved these changes Jan 2, 2024

View reviewed changes

XinShuYang dismissed tnqn’s stale review via 509d564 January 2, 2024 08:09

XinShuYang force-pushed the windhcp branch from ee8222f to 509d564 Compare January 2, 2024 08:09

tnqn approved these changes Jan 2, 2024

View reviewed changes

tnqn merged commit 923b429 into antrea-io:main Jan 2, 2024
45 of 52 checks passed

This was referenced Jan 3, 2024

Automated cherry pick of #5819: Add DHCP IP Retries in PrepareHNSNetwork #5835

Merged

Automated cherry pick of #5819: Add DHCP IP Retries in PrepareHNSNetwork #5836

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add DHCP IP Retries in PrepareHNSNetwork #5819

Add DHCP IP Retries in PrepareHNSNetwork #5819

XinShuYang commented Dec 21, 2023 •

edited

Loading

XinShuYang commented Dec 21, 2023

XinShuYang commented Dec 21, 2023

XinShuYang commented Dec 21, 2023

wenyingd left a comment

XinShuYang commented Dec 25, 2023

XinShuYang commented Dec 28, 2023

wenyingd left a comment

XinShuYang commented Dec 28, 2023

XinShuYang commented Dec 28, 2023

XinShuYang commented Dec 28, 2023

tnqn Dec 28, 2023

XinShuYang Dec 28, 2023 •

edited

Loading

tnqn Dec 28, 2023

XinShuYang Dec 28, 2023

wenyingd Dec 28, 2023

XinShuYang commented Dec 28, 2023

XinShuYang commented Dec 31, 2023

tnqn Jan 2, 2024

tnqn Jan 2, 2024

tnqn commented Jan 2, 2024

tnqn commented Jan 2, 2024

XinShuYang commented Jan 2, 2024

tnqn commented Jan 2, 2024

	klog.ErrorS(err, "Failed to get Ipv4 DHCP status on the network adapter", "adapter", uplinkAdapter.Name)
	klog.ErrorS(err, "Failed to get IPv4 DHCP status on the network adapter", "adapter", uplinkAdapter.Name)

Add DHCP IP Retries in PrepareHNSNetwork #5819

Add DHCP IP Retries in PrepareHNSNetwork #5819

Conversation

XinShuYang commented Dec 21, 2023 • edited Loading

XinShuYang commented Dec 21, 2023

XinShuYang commented Dec 21, 2023

XinShuYang commented Dec 21, 2023

wenyingd left a comment

Choose a reason for hiding this comment

XinShuYang commented Dec 25, 2023

XinShuYang commented Dec 28, 2023

wenyingd left a comment

Choose a reason for hiding this comment

XinShuYang commented Dec 28, 2023

XinShuYang commented Dec 28, 2023

XinShuYang commented Dec 28, 2023

tnqn Dec 28, 2023

Choose a reason for hiding this comment

XinShuYang Dec 28, 2023 • edited Loading

Choose a reason for hiding this comment

tnqn Dec 28, 2023

Choose a reason for hiding this comment

XinShuYang Dec 28, 2023

Choose a reason for hiding this comment

wenyingd Dec 28, 2023

Choose a reason for hiding this comment

XinShuYang commented Dec 28, 2023

XinShuYang commented Dec 31, 2023

tnqn Jan 2, 2024

Choose a reason for hiding this comment

tnqn Jan 2, 2024

Choose a reason for hiding this comment

tnqn commented Jan 2, 2024

tnqn commented Jan 2, 2024

XinShuYang commented Jan 2, 2024

tnqn commented Jan 2, 2024

XinShuYang commented Dec 21, 2023 •

edited

Loading

XinShuYang Dec 28, 2023 •

edited

Loading