On Azure rhel8.4 VM antrea-agent goes into a state where it cannot manage ExternalNode.

[Anandkumar26]

Describe the bug
When an ExternalNode for a RHEL VM is deleted, there is an error in antrea-agent while restoring interface configuration back to uplink. Because of the nephe-ci breaks, where each test adds ExternalNode, performs NetworkPolicy operation and then deletes ExternalNode.

To Reproduce
Spin up RHEL 8.4 VM on Azure.
Create an ExternalNode of RHEL VM, and install antrea-agent on VM.
Delete the ExternalNode, check antrea-agent logs (there should not be any error while restoring the interface config)
Repeat this step, to reproduce the issue. (Usually on the 2nd try we hit the issue)

Expected
There should not be any error while restoring interface configuration on the uplink.

Actual behavior
Obseved an error while restoring configuration on the uplink. As a consequence of it, when we re-add the ExternalNode, antrea-agent will not be able to process the ExternalNode ADD event.

Versions:
Antrea version: v1.12.0

Additional context
antrea-agent logs:

I0605 08:34:22.346209 1 reconciler.go:990] Releasing stale priority 14800
I0605 08:34:22.348248 1 reconciler.go:990] Releasing stale priority 14800
I0605 08:34:22.348278 1 reconciler.go:983] Uninstalling ofRule 2
I0605 08:34:22.348284 1 network_policy.go:1594] ofPriority 14900 is now stale
I0605 08:34:22.352550 1 reconciler.go:990] Releasing stale priority 14900
I0605 08:34:22.363920 1 net_linux.go:283] "Renaming interface" oldName="eth0~" newName="eth0"
I0605 08:34:22.476222 1 external_node_controller.go:601] "Recovered uplink name to the host interface name" uplinkIfName="eth0~" hostInterface="eth0"
I0605 08:34:22.476356 1 net_linux.go:175] "Expected addresses" count=2 addresses=[10.110.0.5/24 fe80::20d:3aff:fe36:cdde/64]
I0605 08:34:22.476434 1 net_linux.go:188] "Existing addresses" count=1 addresses=[10.110.0.5/24 eth0]
I0605 08:34:22.476443 1 net_linux.go:206] Adding address fe80::20d:3aff:fe36:cdde/64 to interface eth0
E0605 08:34:22.476550 1 net_linux.go:330] "Failed to replace route entry" err="network is unreachable" route="{Ifindex: 2 Dst: Src: Gw: 10.110.0.1 Flags: [] Table: 254 Realm: 0}"
E0605 08:34:22.493773 1 external_node_controller.go:237] "Error syncing ExternalNode" err="network is unreachable" ExternalNode="vm-ns/rhel84"
I0605 08:34:27.494580 1 external_node_controller.go:386] "Deleting interface" ifName="eth0"
I0605 08:34:27.494595 1 client.go:490] "Cached flow with provided key was not found" key="eth0"
I0605 08:34:27.494603 1 external_node_controller.go:558] "Removed the flows installed to forward packet between uplinkPort and hostPort" hostInterface="eth0"
I0605 08:34:27.495272 1 external_node_controller.go:577] "Deleted host port in OVS" hostInterface="eth0"
I0605 08:34:27.495433 1 external_node_controller.go:581] "Deleted uplink port in OVS" uplinkIfName="eth0~"
E0605 08:34:29.501705 1 external_node_controller.go:237] "Error syncing ExternalNode" err="failed to wait for host interface eth0 deletion in 2s, err timed out waiting for the condition" ExternalNode="vm-ns/rhel84"
I0605 08:34:39.502354 1 external_node_controller.go:386] "Deleting interface" ifName="eth0"
I0605 08:34:39.502369 1 client.go:490] "Cached flow with provided key was not found" key="eth0"
I0605 08:34:39.502376 1 external_node_controller.go:558] "Removed the flows installed to forward packet between uplinkPort and hostPort" hostInterface="eth0"

[wenyingd]

I created another issue (#5192) to track the dhclient unexpected behavior.