clean linkstatus on node remove and fix vlanconfig validation error

[yaocw2020]

Problem:

1. Clear link statuses of the deleted node in all the link monitor
2. Skip vlanconfig webhook(includes validator and mutator) if the spec is not changed.

Solution:

1. Add OnRemove function in the node controller to clear link statuses
2. Use reflect to compare the spec of the old vlanconfig and new vlanconfig. If it's not changed, skip the webhook.

By the way, use a more elegant way to upgrade golang.org/x/net

Related Issue:
harvester/harvester#3685
harvester/harvester#3697

Test plan:

Case 1:

• Set up a cluster with one node which has more than one NIC for the custom cluster network.
• Add a new cluster network and network config which selects all nodes.
• Join a new node whose NICs are the same as the first node. It would be added successfully and the network config also takes effect. The issue [BUG] bonded NIC cluster network doesn't get added to new node if VMs are up harvester#3685 is validated.
• Execute kubectl get lms -o yaml to check link statuses of all nodes.
• Delete the second node. Check link statuses again and the link statuses of the removed node are deleted. The issue [BUG] Clear network resources when the node is removed harvester#3697 is validated.

Case2:

• Set up a cluster with two nodes. The compute node has more than one NIC.
• Add a new cluster network and network config which selects the compute node.
• Delete the compute node. The cluster network will be unready. The value of annotation network.harvesterhci.io/matched-nodes is "[]". The related vlanstatus is deleted.

[futuretea]

Maybe this PR should be divided into two, the validation of harvester/harvester#3685 passed in my env

[futuretea]

LGTM, Thanks! Test PASS

❯ kubectl -n harvester-system get po -o custom-columns='NAME:metadata.name,IMAGES:spec.containers[*].image'
NAME                                                    IMAGES
harvester-7bc4cd4ff8-7hm8t                              rancher/harvester:v1.1-head
harvester-load-balancer-6f97f675f9-qtcrz                rancher/harvester-load-balancer:v0.1.4
harvester-network-controller-gwzz8                      harbor.futuretea.me/rancher/harvester-network-controller:c00dd1a-amd64
harvester-network-controller-manager-6746b59c75-4vlqc   harbor.futuretea.me/rancher/harvester-network-controller:c00dd1a-amd64
harvester-network-webhook-5b999f8649-gn7zf              harbor.futuretea.me/rancher/harvester-network-webhook:c00dd1a-amd64
harvester-node-disk-manager-njjkt                       rancher/harvester-node-disk-manager:v0.4.10
harvester-node-manager-4svxh                            rancher/harvester-node-manager:v0.1.3
harvester-webhook-79955cff6b-67cp2                      rancher/harvester-webhook:v1.1-head
kube-vip-4lv6x                                          ghcr.io/kube-vip/kube-vip:v0.5.10
kube-vip-cloud-provider-0                               kubevip/kube-vip-cloud-provider:v0.0.1
virt-api-7cfd78b999-rwmcn                               registry.suse.com/suse/sles/15.4/virt-api:0.54.0-150400.3.7.1
virt-controller-5d54b8b9bf-6m8fd                        registry.suse.com/suse/sles/15.4/virt-controller:0.54.0-150400.3.7.1
virt-controller-5d54b8b9bf-6qvxz                        registry.suse.com/suse/sles/15.4/virt-controller:0.54.0-150400.3.7.1
virt-handler-nj2zb                                      registry.suse.com/suse/sles/15.4/virt-handler:0.54.0-150400.3.7.1
virt-operator-7bdb8d7bff-hvkhk                          registry.suse.com/suse/sles/15.4/virt-operator:0.54.0-150400.3.7.1

❯ k get lms -oyaml
apiVersion: v1
items:
- apiVersion: network.harvesterhci.io/v1beta1
  kind: LinkMonitor
  metadata:
    creationTimestamp: "2023-03-23T08:49:59Z"
    finalizers:
    - wrangler.cattle.io/harvester-network-link-monitor-controller
    generation: 22
    name: mgmt
    resourceVersion: "147726"
    uid: be0c796c-ea18-4c9b-b682-2fce3b443c5e
  spec:
    targetLinkRule:
      nameRule: mgmt(-br|-bo)
  status:
    linkStatus:
      harv-11-1:
      - index: 4
        mac: "50:50:50:50:50:37"
        name: mgmt-br
        state: up
        type: bridge
      - index: 5
        mac: "50:50:50:50:50:37"
        masterIndex: 4
        name: mgmt-bo
        state: up
        type: bond
- apiVersion: network.harvesterhci.io/v1beta1
  kind: LinkMonitor
  metadata:
    creationTimestamp: "2023-03-23T08:49:58Z"
    finalizers:
    - wrangler.cattle.io/harvester-network-link-monitor-controller
    generation: 55
    name: nic
    resourceVersion: "147721"
    uid: 93eb506a-8681-4ca2-a2ee-9022570b7787
  spec:
    targetLinkRule:
      typeRule: device
  status:
    linkStatus:
      harv-11-1:
      - index: 2
        mac: "50:50:50:50:50:37"
        masterIndex: 5
        name: ens33
        state: up
        type: device
      - index: 3
        mac: 00:0c:29:38:4f:5e
        masterIndex: 77
        name: ens34
        state: up
        type: device
- apiVersion: network.harvesterhci.io/v1beta1
  kind: LinkMonitor
  metadata:
    creationTimestamp: "2023-03-23T08:56:54Z"
    finalizers:
    - wrangler.cattle.io/harvester-network-link-monitor-controller
    generation: 65
    name: test
    resourceVersion: "147724"
    uid: 87c3e38a-eb22-4945-a525-c7e4f5a96a62
  spec:
    targetLinkRule:
      nameRule: test(-br|-bo)
  status:
    linkStatus:
      harv-11-1:
      - index: 10
        mac: be:56:d7:70:a0:a0
        name: test-br
        promiscuous: true
        state: up
        type: bridge
      - index: 77
        mac: 00:0c:29:38:4f:5e
        masterIndex: 10
        name: test-bo
        state: up
        type: bond
kind: List
metadata:
  resourceVersion: ""

[guangbochen]

LGTM, and verified that both linkstatus update and vlanconfig validation are working fine.
But have spotted a new error of, could be related to the NAD webhook validation:

time="2023-03-23T14:09:03Z" level=error msg="could not update nad default/test because it's still used by VM(s) default/vm1 which must be stopped at first"
error: stream error: stream ID 3; INTERNAL_ERROR; received from peer

spotted new error

[yaocw2020]

The ready status is wrong before. However it's not used. We fix it together.

[bk201]

lgtm, adding node and deleting node works.

[guangbochen]

LGTM, thanks.