Flannel VxLAN DR 模式-编程知识

Flannel VxLAN DR 模式

news/2024/9/21 16:31:32/文章来源:https://www.cnblogs.com/evescn/p/18369788

Flannel VxLAN DR 模式

一、环境信息

主机	IP
ubuntu	172.16.94.141

软件	版本
docker	26.1.4
helm	v3.15.0-rc.2
kind	0.18.0
clab	0.54.2
kubernetes	1.23.4
ubuntu os	Ubuntu 20.04.6 LTS
kernel	5.11.5 内核升级文档

二、安装服务

`kind` 配置文件信息

$ cat install.sh#!/bin/bash
date
set -v# 1.prep noCNI env
cat <<EOF | kind create cluster --name=clab-flannel-vxlan-directrouting --image=kindest/node:v1.23.4 --config=-
kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
networking:disableDefaultCNI: truepodSubnet: "10.244.0.0/16"
nodes:
- role: control-planekubeadmConfigPatches:- |kind: InitConfigurationnodeRegistration:kubeletExtraArgs:node-ip: 10.1.5.10node-labels: "rack=rack0"- role: workerkubeadmConfigPatches:- |kind: JoinConfigurationnodeRegistration:kubeletExtraArgs:node-ip: 10.1.5.11node-labels: "rack=rack0"- role: workerkubeadmConfigPatches:- |kind: JoinConfigurationnodeRegistration:kubeletExtraArgs:node-ip: 10.1.8.10node-labels: "rack=rack1"- role: workerkubeadmConfigPatches:- |kind: JoinConfigurationnodeRegistration:kubeletExtraArgs:node-ip: 10.1.8.11node-labels: "rack=rack1"containerdConfigPatches:
- |-[plugins."io.containerd.grpc.v1.cri".registry.mirrors."harbor.dayuan1997.com"]endpoint = ["https://harbor.dayuan1997.com"]
EOF# 2.remove taints
controller_node=`kubectl get nodes --no-headers  -o custom-columns=NAME:.metadata.name| grep control-plane`
kubectl taint nodes $controller_node node-role.kubernetes.io/master:NoSchedule-
kubectl get nodes -o wide# 3.install necessary tools
# cd /opt/
# curl -o calicoctl -O -L "https://gh.api.99988866.xyz/https://github.com/containernetworking/plugins/releases/download/v0.9.0/cni-plugins-linux-amd64-v0.9.0.tgz" 
# tar -zxvf cni-plugins-linux-amd64-v0.9.0.tgzfor i in $(docker ps -a --format "table {{.Names}}" | grep flannel) 
doecho $idocker cp /opt/bridge $i:/opt/cni/bin/docker cp /usr/bin/ping $i:/usr/bin/pingdocker exec -it $i bash -c "sed -i -e 's/jp.archive.ubuntu.com\|archive.ubuntu.com\|security.ubuntu.com/old-releases.ubuntu.com/g' /etc/apt/sources.list"docker exec -it $i bash -c "apt-get -y update >/dev/null && apt-get -y install net-tools tcpdump lrzsz bridge-utils >/dev/null 2>&1"
done

安装 k8s 集群

root@kind:~# ./install.shCreating cluster "clab-flannel-vxlan-directrouting" ...✓ Ensuring node image (kindest/node:v1.23.4) 🖼✓ Preparing nodes 📦 📦 📦 📦  ✓ Writing configuration 📜 ✓ Starting control-plane 🕹️ ✓ Installing StorageClass 💾 ✓ Joining worker nodes 🚜 
Set kubectl context to "kind-clab-flannel-vxlan-directrouting"
You can now use your cluster with:kubectl cluster-info --context kind-clab-flannel-vxlan-directroutingHave a nice day! 👋

root@kind:~# kubectl get node -o wide
NAME                                             STATUS     ROLES                  AGE     VERSION   INTERNAL-IP   EXTERNAL-IP   OS-IMAGE       KERNEL-VERSION          CONTAINER-RUNTIME
clab-flannel-vxlan-directrouting-control-plane   NotReady   control-plane,master   2m48s   v1.23.4   <none>        <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker          NotReady   <none>                 2m14s   v1.23.4   <none>        <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker2         NotReady   <none>                 2m14s   v1.23.4   <none>        <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker3         NotReady   <none>                 2m14s   v1.23.4   <none>        <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10

创建 `clab` 容器环境

创建网桥

root@kind:~# brctl addbr br-leaf0
root@kind:~# ifconfig br-leaf0 up
root@kind:~# brctl addbr br-leaf1
root@kind:~# ifconfig br-leaf1 uproot@kind:~# ip a l
19: br-pool0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9500 qdisc noqueue state UP group default qlen 1000link/ether aa:c1:ab:14:8f:99 brd ff:ff:ff:ff:ff:ffinet6 fe80::e8df:fcff:fed4:3e17/64 scope link valid_lft forever preferred_lft forever
20: br-pool1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9500 qdisc noqueue state UP group default qlen 1000link/ether aa:c1:ab:08:cc:9d brd ff:ff:ff:ff:ff:ffinet6 fe80::88c:adff:fef2:f336/64 scope link valid_lft forever preferred_lft forever

创建这两个网桥主要是为了让 kind 上节点通过虚拟交换机连接到 containerLab ，为什么不直连接 containerLab ，如果 10.1.5.10/24 使用 vethPair 和 containerLab 进行连接， 10.1.5.11/24 就没有额外的端口进行连接

`clab` 网络拓扑文件

# flannel.vxlan.directrouting.clab.yml
name: flannel-vxlan-directrouting
topology:nodes:gw0:kind: linuximage: vyos/vyos:1.2.8cmd: /sbin/initbinds:- /lib/modules:/lib/modules- ./startup-conf/gw0-boot.cfg:/opt/vyatta/etc/config/config.bootbr-pool0:kind: bridgebr-pool1:kind: bridgeserver1:kind: linuximage: harbor.dayuan1997.com/devops/nettool:0.9# 复用节点网络，共享网络命名空间network-mode: container:clab-flannel-vxlan-directrouting-control-plane# 配置是为了设置节点上的业务网卡，同时将默认路由的网关进行更改，使用业务网卡为出接口。exec:      - ip addr add 10.1.5.10/24 dev net0      - ip route replace default via 10.1.5.1server2:kind: linuximage: harbor.dayuan1997.com/devops/nettool:0.9# 复用节点网络，共享网络命名空间network-mode: container:clab-flannel-vxlan-directrouting-worker# 配置是为了设置节点上的业务网卡，同时将默认路由的网关进行更改，使用业务网卡为出接口。exec:- ip addr add 10.1.5.11/24 dev net0- ip route replace default via 10.1.5.1server3:kind: linuximage: harbor.dayuan1997.com/devops/nettool:0.9# 复用节点网络，共享网络命名空间network-mode: container:clab-flannel-vxlan-directrouting-worker2# 配置是为了设置节点上的业务网卡，同时将默认路由的网关进行更改，使用业务网卡为出接口。exec:- ip addr add 10.1.8.10/24 dev net0- ip route replace default via 10.1.8.1server4:kind: linuximage: harbor.dayuan1997.com/devops/nettool:0.9# 复用节点网络，共享网络命名空间network-mode: container:clab-flannel-vxlan-directrouting-worker3# 配置是为了设置节点上的业务网卡，同时将默认路由的网关进行更改，使用业务网卡为出接口。exec:- ip addr add 10.1.8.11/24 dev net0- ip route replace default via 10.1.8.1links:- endpoints: ["br-pool0:br-pool0-net0", "server1:net0"]- endpoints: ["br-pool0:br-pool0-net1", "server2:net0"]- endpoints: ["br-pool1:br-pool1-net0", "server3:net0"]- endpoints: ["br-pool1:br-pool1-net1", "server4:net0"]- endpoints: ["gw0:eth1", "br-pool0:br-pool0-net2"]- endpoints: ["gw0:eth2", "br-pool1:br-pool1-net2"]

`VyOS` 配置文件

gw0-boot.cfg

配置文件

# ./startup-conf/gw0-boot.cfg
interfaces {ethernet eth1 {address 10.1.5.1/24duplex autosmp-affinity autospeed auto}ethernet eth2 {address 10.1.8.1/24duplex autosmp-affinity autospeed auto}loopback lo {}
}
# 配置 nat 信息，gw0 网络下的其他服务器可以访问外网
nat {source {rule 100 {outbound-interface eth0source {address 10.1.0.0/16}translation {address masquerade}}}
}
system {config-management {commit-revisions 100}console {device ttyS0 {speed 9600}}host-name vyoslogin {user vyos {authentication {encrypted-password $6$QxPS.uk6mfo$9QBSo8u1FkH16gMyAVhus6fU3LOzvLR9Z9.82m3tiHFAxTtIkhaZSWssSgzt4v4dGAL8rhVQxTg0oAG9/q11h/plaintext-password ""}level admin}}ntp {server 0.pool.ntp.org {}server 1.pool.ntp.org {}server 2.pool.ntp.org {}}syslog {global {facility all {level info}facility protocols {level debug}}}time-zone UTC
}/* Warning: Do not remove the following line. */
/* === vyatta-config-version: "wanloadbalance@3:l2tp@1:pptp@1:ntp@1:mdns@1:webgui@1:conntrack@1:ipsec@5:cluster@1:dhcp-server@5:nat@4:dhcp-relay@2:webproxy@1:system@10:pppoe-server@2:dns-forwarding@1:ssh@1:quagga@7:broadcast-relay@1:qos@1:snmp@1:firewall@5:zone-policy@1:config-management@1:webproxy@2:vrrp@2:conntrack-sync@1" === */
/* Release version: 1.2.8 */

部署服务

# tree -L 2 ./
./
├── flannel.vxlan.directrouting.clab.yml
└── startup-conf└── gw0-boot.cfg# clab deploy -t flannel.vxlan.directrouting.clab.yml

INFO[0000] Containerlab v0.54.2 started                 
INFO[0000] Parsing & checking topology file: clab.yaml  
INFO[0000] Creating docker network: Name="clab", IPv4Subnet="172.20.20.0/24", IPv6Subnet="2001:172:20:20::/64", MTU=1500 
INFO[0000] Creating lab directory: /root/wcni-kind/flannel/4-flannel-vxlan-directrouting/clab-flannel-vxlan-directrouting 
WARN[0000] node clab-flannel-vxlan-directrouting-control-plane referenced in namespace sharing not found in topology definition, considering it an external dependency. 
WARN[0000] node clab-flannel-vxlan-directrouting-worker referenced in namespace sharing not found in topology definition, considering it an external dependency. 
WARN[0000] node clab-flannel-vxlan-directrouting-worker2 referenced in namespace sharing not found in topology definition, considering it an external dependency. 
WARN[0000] node clab-flannel-vxlan-directrouting-worker3 referenced in namespace sharing not found in topology definition, considering it an external dependency. 
INFO[0000] Creating container: "gw0"                    
INFO[0001] Created link: gw0:eth1 <--> br-pool0:br-pool0-net2 
INFO[0001] Created link: gw0:eth2 <--> br-pool1:br-pool1-net2 
INFO[0003] Creating container: "server2"                
INFO[0003] Creating container: "server1"                
INFO[0003] Created link: br-pool0:br-pool0-net1 <--> server2:net0 
INFO[0003] Created link: br-pool0:br-pool0-net0 <--> server1:net0 
INFO[0004] Creating container: "server3"                
INFO[0004] Creating container: "server4"                
INFO[0004] Created link: br-pool1:br-pool1-net0 <--> server3:net0 
INFO[0004] Created link: br-pool1:br-pool1-net1 <--> server4:net0 
INFO[0005] Executed command "ip addr add 10.1.5.10/24 dev net0" on the node "server1". stdout: 
INFO[0005] Executed command "ip route replace default via 10.1.5.1" on the node "server1". stdout: 
INFO[0005] Executed command "ip addr add 10.1.8.11/24 dev net0" on the node "server4". stdout: 
INFO[0005] Executed command "ip route replace default via 10.1.8.1" on the node "server4". stdout: 
INFO[0005] Executed command "ip addr add 10.1.8.10/24 dev net0" on the node "server3". stdout: 
INFO[0005] Executed command "ip route replace default via 10.1.8.1" on the node "server3". stdout: 
INFO[0005] Executed command "ip addr add 10.1.5.11/24 dev net0" on the node "server2". stdout: 
INFO[0005] Executed command "ip route replace default via 10.1.5.1" on the node "server2". stdout: 
INFO[0005] Adding containerlab host entries to /etc/hosts file 
INFO[0005] Adding ssh config for containerlab nodes     
INFO[0005] 🎉 New containerlab version 0.56.0 is available! Release notes: https://containerlab.dev/rn/0.56/
Run 'containerlab version upgrade' to upgrade or go check other installation options at https://containerlab.dev/install/ 
+---+------------------------------------------+--------------+------------------------------------------+-------+---------+----------------+----------------------+
| # |                   Name                   | Container ID |                  Image                   | Kind  |  State  |  IPv4 Address  |     IPv6 Address     |
+---+------------------------------------------+--------------+------------------------------------------+-------+---------+----------------+----------------------+
| 1 | clab-flannel-vxlan-directrouting-gw0     | 2ac429824caa | vyos/vyos:1.2.8                          | linux | running | 172.20.20.2/24 | 2001:172:20:20::2/64 |
| 2 | clab-flannel-vxlan-directrouting-server1 | c3cc494ff542 | harbor.dayuan1997.com/devops/nettool:0.9 | linux | running | N/A            | N/A                  |
| 3 | clab-flannel-vxlan-directrouting-server2 | bc24767595b4 | harbor.dayuan1997.com/devops/nettool:0.9 | linux | running | N/A            | N/A                  |
| 4 | clab-flannel-vxlan-directrouting-server3 | 71c0daf892e3 | harbor.dayuan1997.com/devops/nettool:0.9 | linux | running | N/A            | N/A                  |
| 5 | clab-flannel-vxlan-directrouting-server4 | b8361f60cfe6 | harbor.dayuan1997.com/devops/nettool:0.9 | linux | running | N/A            | N/A                  |
+---+------------------------------------------+--------------+------------------------------------------+-------+---------+----------------+----------------------+

检查 k8s 集群信息

root@kind:~# kubectl get node -o wide
NAME                                             STATUS     ROLES                  AGE     VERSION   INTERNAL-IP   EXTERNAL-IP   OS-IMAGE       KERNEL-VERSION          CONTAINER-RUNTIME
clab-flannel-vxlan-directrouting-control-plane   NotReady   control-plane,master   8m38s   v1.23.4   10.1.5.10     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker          NotReady   <none>                 8m4s    v1.23.4   10.1.5.11     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker2         NotReady   <none>                 8m4s    v1.23.4   10.1.8.10     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker3         NotReady   <none>                 8m4s    v1.23.4   10.1.8.11     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10# 查看 node 节点 ip 信息
root@kind:~# docker exec -it clab-flannel-vxlan-directrouting-control-plane ip a l
17: eth0@if18: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default link/ether 02:42:ac:12:00:05 brd ff:ff:ff:ff:ff:ff link-netnsid 0inet 172.18.0.5/16 brd 172.18.255.255 scope global eth0valid_lft forever preferred_lft foreverinet6 fc00:f853:ccd:e793::5/64 scope global nodad valid_lft forever preferred_lft foreverinet6 fe80::42:acff:fe12:5/64 scope link valid_lft forever preferred_lft forever
22: net0@if23: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9500 qdisc noqueue state UP group default link/ether aa:c1:ab:ab:fa:3b brd ff:ff:ff:ff:ff:ff link-netnsid 0inet 10.1.5.10/24 scope global net0valid_lft forever preferred_lft foreverinet6 fe80::a8c1:abff:feab:fa3b/64 scope link valid_lft forever preferred_lft forever# 查看 node 节点路由信息
root@kind:~# docker exec -it clab-flannel-vxlan-directrouting-control-plane ip r s
default via 10.1.5.1 dev net0 
10.1.5.0/24 dev net0 proto kernel scope link src 10.1.5.10 
172.18.0.0/16 dev eth0 proto kernel scope link src 172.18.0.5

查看 k8s 集群发现 node 节点 ip 地址分配了，登陆容器查看到了新的 ip 地址，并且默认路由信息调整为了 10.1.5.0/24 dev net0 proto kernel scope link src 10.1.5.10

安装 `flannel` 服务

flannel.yaml

配置文件

# flannel.yaml
---
kind: Namespace
apiVersion: v1
metadata:name: kube-flannellabels:pod-security.kubernetes.io/enforce: privileged
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:name: flannel
rules:
- apiGroups:- ""resources:- podsverbs:- get
- apiGroups:- ""resources:- nodesverbs:- list- watch
- apiGroups:- ""resources:- nodes/statusverbs:- patch
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:name: flannel
roleRef:apiGroup: rbac.authorization.k8s.iokind: ClusterRolename: flannel
subjects:
- kind: ServiceAccountname: flannelnamespace: kube-system
---
apiVersion: v1
kind: ServiceAccount
metadata:name: flannelnamespace: kube-system
---
kind: ConfigMap
apiVersion: v1
metadata:name: kube-flannel-cfgnamespace: kube-systemlabels:tier: nodeapp: flannel
data:cni-conf.json: |{"name": "cbr0","cniVersion": "0.3.1","plugins": [{"type": "flannel","delegate": {"hairpinMode": true,"isDefaultGateway": true}},{"type": "portmap","capabilities": {"portMappings": true}}]}net-conf.json: |{"Network": "10.244.0.0/16","Backend": {"Type": "vxlan","DirectRouting": true}}
---
apiVersion: apps/v1
kind: DaemonSet
metadata:name: kube-flannel-dsnamespace: kube-systemlabels:tier: nodeapp: flannel
spec:selector:matchLabels:app: flanneltemplate:metadata:labels:tier: nodeapp: flannelspec:affinity:nodeAffinity:requiredDuringSchedulingIgnoredDuringExecution:nodeSelectorTerms:- matchExpressions:- key: kubernetes.io/osoperator: Invalues:- linuxhostNetwork: truepriorityClassName: system-node-criticaltolerations:- operator: Existseffect: NoScheduleserviceAccountName: flannelinitContainers:- name: install-cni-plugin#image: flannelcni/flannel-cni-plugin:v1.1.0 for ppc64le and mips64le (dockerhub limitations may apply)#image: 192.168.2.100:5000/rancher/mirrored-flannelcni-flannel-cni-plugin:v1.1.0image: harbor.dayuan1997.com/devops/rancher/mirrored-flannelcni-flannel-cni-plugin:v1.1.0command:- cpargs:- -f- /flannel- /opt/cni/bin/flannelvolumeMounts:- name: cni-pluginmountPath: /opt/cni/bin- name: install-cni#image: flannelcni/flannel:v0.19.2 for ppc64le and mips64le (dockerhub limitations may apply)#image: 192.168.2.100:5000/rancher/mirrored-flannelcni-flannel:v0.19.2image: harbor.dayuan1997.com/devops/rancher/mirrored-flannelcni-flannel:v0.19.2command:- cpargs:- -f- /etc/kube-flannel/cni-conf.json- /etc/cni/net.d/10-flannel.conflistvolumeMounts:- name: cnimountPath: /etc/cni/net.d- name: flannel-cfgmountPath: /etc/kube-flannel/containers:- name: kube-flannel#image: flannelcni/flannel:v0.19.2 for ppc64le and mips64le (dockerhub limitations may apply)#image: 192.168.2.100:5000/rancher/mirrored-flannelcni-flannel:v0.19.2image: harbor.dayuan1997.com/devops/rancher/mirrored-flannelcni-flannel:v0.19.2command:- /opt/bin/flanneldargs:- --ip-masq- --kube-subnet-mgrresources:requests:cpu: "100m"memory: "50Mi"limits:cpu: "100m"memory: "50Mi"securityContext:privileged: falsecapabilities:add: ["NET_ADMIN", "NET_RAW"]env:- name: POD_NAMEvalueFrom:fieldRef:fieldPath: metadata.name- name: POD_NAMESPACEvalueFrom:fieldRef:fieldPath: metadata.namespace- name: EVENT_QUEUE_DEPTHvalue: "5000"volumeMounts:- name: runmountPath: /run/flannel- name: flannel-cfgmountPath: /etc/kube-flannel/- name: xtables-lockmountPath: /run/xtables.lock- name: tunmountPath: /dev/net/tunvolumes:- name: tunhostPath:path: /dev/net/tun- name: runhostPath:path: /run/flannel- name: cni-pluginhostPath:path: /opt/cni/bin- name: cnihostPath:path: /etc/cni/net.d- name: flannel-cfgconfigMap:name: kube-flannel-cfg- name: xtables-lockhostPath:path: /run/xtables.locktype: FileOrCreate

flannel.yaml 参数解释

Backend.Type
- 含义: 用于指定 flannel 工作模式。
- vxlan: flannel 工作在 vxlan 模式。
Backend.DirectRouting
- 含义: 用于指定 vxlan 模式同网段使用 host-gw 模式。
- true: vxlan 模式中，同网段的 node 节点之间数据转发使用 host-gw 模式。

root@kind:~# kubectl apply -f flannel.yaml
namespace/kube-flannel created
clusterrole.rbac.authorization.k8s.io/flannel created
clusterrolebinding.rbac.authorization.k8s.io/flannel created
serviceaccount/flannel created
configmap/kube-flannel-cfg created
daemonset.apps/kube-flannel-ds created

查看 k8s 集群和 flannel 服务

root@kind:~# kubectl get node -o wide
NAME                                             STATUS   ROLES                  AGE   VERSION   INTERNAL-IP   EXTERNAL-IP   OS-IMAGE       KERNEL-VERSION          CONTAINER-RUNTIME
clab-flannel-vxlan-directrouting-control-plane   Ready    control-plane,master   13m   v1.23.4   10.1.5.10     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker          Ready    <none>                 13m   v1.23.4   10.1.5.11     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker2         Ready    <none>                 13m   v1.23.4   10.1.8.10     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10
clab-flannel-vxlan-directrouting-worker3         Ready    <none>                 13m   v1.23.4   10.1.8.11     <none>        Ubuntu 21.10   5.11.5-051105-generic   containerd://1.5.10

查看安装的服务

root@kind:~# kubectl get pods -A
NAMESPACE            NAME                                                                     READY   STATUS    RESTARTS   AGE
kube-system          coredns-64897985d-8dt79                                                  1/1     Running   0          13m
kube-system          coredns-64897985d-sprng                                                  1/1     Running   0          13m
kube-system          etcd-clab-flannel-vxlan-directrouting-control-plane                      1/1     Running   0          14m
kube-system          kube-apiserver-clab-flannel-vxlan-directrouting-control-plane            1/1     Running   0          14m
kube-system          kube-controller-manager-clab-flannel-vxlan-directrouting-control-plane   1/1     Running   0          14m
kube-system          kube-flannel-ds-6b4sk                                                    1/1     Running   0          3m52s
kube-system          kube-flannel-ds-6vqf2                                                    1/1     Running   0          3m52s
kube-system          kube-flannel-ds-76czr                                                    1/1     Running   0          3m52s
kube-system          kube-flannel-ds-xhw4f                                                    1/1     Running   0          3m52s
kube-system          kube-proxy-d9kjr                                                         1/1     Running   0          13m
kube-system          kube-proxy-fl2v7                                                         1/1     Running   0          13m
kube-system          kube-proxy-mgv2m                                                         1/1     Running   0          13m
kube-system          kube-proxy-xcl4n                                                         1/1     Running   0          13m
kube-system          kube-scheduler-clab-flannel-vxlan-directrouting-control-plane            1/1     Running   0          14m
local-path-storage   local-path-provisioner-5ddd94ff66-lk6kc                                  1/1     Running   0          13m

`k8s` 集群安装 `Pod` 测试网络

root@kind:~# cat cni.yamlapiVersion: apps/v1
kind: DaemonSet
#kind: Deployment
metadata:labels:app: cniname: cni
spec:#replicas: 1selector:matchLabels:app: cnitemplate:metadata:labels:app: cnispec:containers:- image: harbor.dayuan1997.com/devops/nettool:0.9name: nettoolboxsecurityContext:privileged: true---
apiVersion: v1
kind: Service
metadata:name: serversvc
spec:type: NodePortselector:app: cniports:- name: cniport: 80targetPort: 80nodePort: 32000

root@kind:~# kubectl apply -f cni.yaml
daemonset.apps/cni created
service/serversvc createdroot@kind:~# kubectl run net --image=harbor.dayuan1997.com/devops/nettool:0.9
pod/net created

查看安装服务信息

root@kind:~# kubectl get pods -o wide
NAME        READY   STATUS    RESTARTS   AGE   IP           NODE                                             NOMINATED NODE   READINESS GATES
cni-5sxkl   1/1     Running   0          18s   10.244.2.2   clab-flannel-vxlan-directrouting-worker2         <none>           <none>
cni-vwgqx   1/1     Running   0          18s   10.244.0.2   clab-flannel-vxlan-directrouting-control-plane   <none>           <none>
cni-w79ph   1/1     Running   0          18s   10.244.3.5   clab-flannel-vxlan-directrouting-worker3         <none>           <none>
cni-x2rxr   1/1     Running   0          18s   10.244.2.2   clab-flannel-vxlan-directrouting-worker          <none>           <none>
net         1/1     Running   0          13s   10.244.1.3   clab-flannel-vxlan-directrouting-worker2         <none>           <none>root@kind:~# kubectl get svc 
NAME         TYPE        CLUSTER-IP    EXTERNAL-IP   PORT(S)        AGE
kubernetes   ClusterIP   10.96.0.1     <none>        443/TCP        15m
serversvc    NodePort    10.96.50.15   <none>        80:32000/TCP   27s

三、测试网络

同节点 `Pod` 网络通讯

可以查看此文档 Flannel UDP 模式中，同节点网络通讯，数据包转发流程一致

Flannel 同节点通信通过 l2 网络， 2 层交换机完成

跨节点同 `Node` 网段 `Pod` 网络通讯

可以查看此文档 Flannel HOST-GW 模式中，不同节点 Pod 网络通讯，数据包转发流程一致

跨节点不同 `Node` 网段 `Pod` 网络通讯

Pod 节点信息

## ip 信息
root@kind:~# kubectl exec -it net -- ip a l
3: eth0@if5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9450 qdisc noqueue state UP group default link/ether 26:a8:8d:74:59:3f brd ff:ff:ff:ff:ff:ff link-netnsid 0inet 10.244.1.3/24 brd 10.244.2.255 scope global eth0valid_lft forever preferred_lft foreverinet6 fe80::24a8:8dff:fe74:593f/64 scope link valid_lft forever preferred_lft forever## 路由信息
root@kind:~# kubectl exec -it net -- ip r s
default via 10.244.1.1 dev eth0 
10.244.0.0/16 via 10.244.1.1 dev eth0 
10.244.1.0/24 dev eth0 proto kernel scope link src 10.244.1.3

Pod 节点所在 Node 节点信息

root@kind:~# docker exec -it clab-flannel-vxlan-directrouting-worker2 bash## ip 信息
root@clab-flannel-vxlan-directrouting-worker2:/# ip a l 
2: flannel.1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9450 qdisc noqueue state UNKNOWN group default link/ether ca:7d:71:63:e2:a2 brd ff:ff:ff:ff:ff:ffinet 10.244.1.0/32 scope global flannel.1valid_lft forever preferred_lft foreverinet6 fe80::c87d:71ff:fe63:e2a2/64 scope link valid_lft forever preferred_lft forever
3: cni0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9450 qdisc noqueue state UP group default qlen 1000link/ether d2:3c:81:f3:9c:6f brd ff:ff:ff:ff:ff:ffinet 10.244.1.1/24 brd 10.244.2.255 scope global cni0valid_lft forever preferred_lft foreverinet6 fe80::d03c:81ff:fef3:9c6f/64 scope link valid_lft forever preferred_lft forever
4: vethd3fcbe42@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9450 qdisc noqueue master cni0 state UP group default link/ether 92:01:1c:7e:82:65 brd ff:ff:ff:ff:ff:ff link-netns cni-baf1f367-320c-1afa-a624-88ba3fc51a48inet6 fe80::9001:1cff:fe7e:8265/64 scope link valid_lft forever preferred_lft forever
5: veth25e540eb@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9450 qdisc noqueue master cni0 state UP group default link/ether b6:10:4f:07:c1:0e brd ff:ff:ff:ff:ff:ff link-netns cni-07b75239-d571-5b75-5257-e3c3e6ed5c01inet6 fe80::b410:4fff:fe07:c10e/64 scope link valid_lft forever preferred_lft forever
13: eth0@if14: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default link/ether 02:42:ac:12:00:03 brd ff:ff:ff:ff:ff:ff link-netnsid 0inet 172.18.0.3/16 brd 172.18.255.255 scope global eth0valid_lft forever preferred_lft foreverinet6 fc00:f853:ccd:e793::3/64 scope global nodad valid_lft forever preferred_lft foreverinet6 fe80::42:acff:fe12:3/64 scope link valid_lft forever preferred_lft forever
28: net0@if29: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9500 qdisc noqueue state UP group default link/ether aa:c1:ab:84:fd:ec brd ff:ff:ff:ff:ff:ff link-netnsid 0inet 10.1.8.10/24 scope global net0valid_lft forever preferred_lft foreverinet6 fe80::a8c1:abff:fe84:fdec/64 scope link valid_lft forever preferred_lft forever## 路由信息
root@clab-flannel-vxlan-directrouting-worker2:/# ip r s
default via 10.1.8.1 dev net0 
10.1.8.0/24 dev net0 proto kernel scope link src 10.1.8.10 
10.244.0.0/24 via 10.244.0.0 dev flannel.1 onlink 
10.244.1.0/24 dev cni0 proto kernel scope link src 10.244.1.1 
10.244.2.0/24 via 10.244.2.0 dev flannel.1 onlink 
10.244.3.0/24 via 10.1.8.11 dev net0 
172.18.0.0/16 dev eth0 proto kernel scope link src 172.18.0.3

Pod 节点进行 ping 包测试，访问 cni-x2rxr Pod 节点

root@kind:~# kubectl exec -it net -- ping 10.244.2.2 -c 1
PING 10.244.2.2 (10.244.2.2): 56 data bytes
64 bytes from 10.244.2.2: seq=0 ttl=62 time=1.402 ms--- 10.244.2.2 ping statistics ---
1 packets transmitted, 1 packets received, 0% packet loss
round-trip min/avg/max = 1.402/1.402/1.402 ms

Pod 节点 eth0 网卡抓包

net~$ tcpdump -pne -i eth0
08:23:24.189370 26:a8:8d:74:59:3f > d2:3c:81:f3:9c:6f, ethertype IPv4 (0x0800), length 98: 10.244.1.3 > 10.244.2.2: ICMP echo request, id 54, seq 14, length 64
08:23:24.189496 d2:3c:81:f3:9c:6f > 26:a8:8d:74:59:3f, ethertype IPv4 (0x0800), length 98: 10.244.2.2 > 10.244.1.3: ICMP echo reply, id 54, seq 14, length 64

数据包源 mac 地址: 26:a8:8d:74:59:3f 为 eth0 网卡 mac 地址，而目的 mac 地址: d2:3c:81:f3:9c:6f 为 net Pod 节点 cni0 网卡对应的网卡 mac 地址，cni0
网卡 ip 地址为网络网关地址 10.244.2.1 ， flannel 为 2 层网络模式通过路由送往数据到网关地址

net~$ arp -n
Address                  HWtype  HWaddress           Flags Mask            Iface
10.244.1.1               ether   d2:3c:81:f3:9c:6f   C                     eth0

而通过 veth pair 可以确定 Pod 节点 eth0 网卡对应的 veth pair 为 veth25e540eb@if3 网卡

clab-flannel-vxlan-directrouting-worker2 节点 veth25e540eb 网卡抓包

root@clab-flannel-vxlan-directrouting-worker2:/# tcpdump -pne -i veth25e540eb
08:26:28.784300 26:a8:8d:74:59:3f > 42:a5:47:f5:4b:9f, ethertype IPv4 (0x0800), length 98: 10.244.1.3 > 10.244.1.2: ICMP echo request, id 23, seq 1576, length 64
08:26:28.784371 42:a5:47:f5:4b:9f > 26:a8:8d:74:59:3f, ethertype IPv4 (0x0800), length 98: 10.244.1.2 > 10.244.1.3: ICMP echo reply, id 23, seq 1576, length 64

因为他们互为 veth pair 所以抓包信息相同

clab-flannel-vxlan-directrouting-worker2 节点 cni0 网卡抓包

root@clab-flannel-vxlan-directrouting-worker2:/# tcpdump -pne -i cni0
08:26:45.256754 26:a8:8d:74:59:3f > d2:3c:81:f3:9c:6f, ethertype IPv4 (0x0800), length 98: 10.244.1.3 > 10.244.2.2: ICMP echo request, id 54, seq 215, length 64
08:26:45.256878 d2:3c:81:f3:9c:6f > 26:a8:8d:74:59:3f, ethertype IPv4 (0x0800), length 98: 10.244.2.2 > 10.244.1.3: ICMP echo reply, id 54, seq 215, length 64

数据包源 mac 地址: 26:a8:8d:74:59:3f 为 net Pod 节点 eth0 网卡 mac 地址，而目的 mac 地址: d2:3c:81:f3:9c:6f 为 cni0 网卡 mac 地址

查看 clab-flannel-vxlan-directrouting-worker2 主机路由信息，发现并在数据包会在通过 10.244.2.0/24 via 10.244.2.0 dev flannel.1 onlink 路由信息转发

clab-flannel-vxlan-directrouting-worker2 节点 flannel.1 网卡抓包

root@clab-flannel-vxlan-directrouting-worker2:/# tcpdump -pne -i flannel.1 icmp
08:30:06.337766 ca:7d:71:63:e2:a2 > 3e:0a:47:72:56:37, ethertype IPv4 (0x0800), length 98: 10.244.1.3 > 10.244.2.2: ICMP echo request, id 54, seq 416, length 64
08:30:06.337895 3e:0a:47:72:56:37 > ca:7d:71:63:e2:a2, ethertype IPv4 (0x0800), length 98: 10.244.2.2 > 10.244.1.3: ICMP echo reply, id 54, seq 416, length 64

数据包源 mac 地址: ca:7d:71:63:e2:a2 为 clab-flannel-vxlan-directrouting-worker2 节点 flannel.1 网卡 mac 地址，而目的 mac 地址: 3e:0a:47:72:56:37 是谁的 mac 地址？查看宿主机 arp 信息，目的 mac 地址: 3e:0a:47:72:56:37 是 10.244.2.0 网段 mac 地址，这个地址如何学习到的？可以查看 FDB 自动学习绑定过程检测

root@clab-flannel-vxlan-directrouting-worker2:/# arp -n
Address                  HWtype  HWaddress           Flags Mask            Iface
10.244.2.0               ether   3e:0a:47:72:56:37   CM                    flannel.1

查看 fdb 信息

root@clab-flannel-vxlan-directrouting-worker2:/# bridge fdb show
3e:0a:47:72:56:37 dev flannel.1 dst 10.1.5.11 self permanent

通过查看 fdb 表信息可以看到 3e:0a:47:72:56:37 dev flannel.1 dst 10.1.5.11 self permanent 标示了 mac 地址 3e:0a:47:72:56:37 所在的主机为 10.1.5.11 ，因此 vxlan 封装的外层数据的目的 ip 是使用 10.1.5.11，而 vxlan 封装的外层数据的源的 ip 是本机 net0 网卡 ip

clab-flannel-vxlan-directrouting-worker2 节点 net0 网卡抓包

request 数据包信息信息
- icmp 包中，外部 mac 信息中，源 mac: aa:c1:ab:84:fd:ec 为 clab-flannel-vxlan-directrouting-worker2 的 net0 网卡 mac ，目的 mac: aa:c1:ab:d2:3d:96 为 gw0 主机的 eth2 网卡 mac。使用 udp 协议 8472 端口进行数据传输，vxlan 信息中 vni 为 1。
- icmp 包中，内部 mac 信息中，源 mac: ca:7d:71:63:e2:a2 为 clab-flannel-vxlan-directrouting-worker2 的 flannel.1 网卡 mac ，目的 mac: 3e:0a:47:72:56:37 为对端 clab-flannel-vxlan-directrouting-worker 主机的 flannel.1 网卡 mac。
clab-flannel-vxlan-directrouting-worker2 节点 vxlan 信息

root@clab-flannel-vxlan-directrouting-worker2:/# ip -d link show
2: flannel.1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9450 qdisc noqueue state UNKNOWN mode DEFAULT group default link/ether ca:7d:71:63:e2:a2 brd ff:ff:ff:ff:ff:ff promiscuity 0 minmtu 68 maxmtu 65535 vxlan id 1 local 10.1.8.10 dev net0 srcport 0 0 dstport 8472 nolearning ttl auto ageing 300 udpcsum noudp6zerocsumtx noudp6zerocsumrx addrgenmode eui64 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535

数据包流向

数据从 pod 服务发出，通过查看本机路由表，送往 10.244.1.1 网卡。路由: 10.244.0.0/16 via 10.244.1.1 dev eth0
通过 veth pair 网卡 veth25e540eb 发送数据到 clab-flannel-vxlan-directrouting-worker2 主机上，在转送到 cni0: 10.244.1.1 网卡
clab-flannel-vxlan-directrouting-worker2 主机查看自身路由后，会送往 flannel.1 接口，因为目的地址为 10.244.2.2。路由: 10.244.2.0/24 via 10.244.2.0 dev flannel.1 onlink
flannel.1 接口为 vxlan 模式，会重新封装数据包，封装信息查看 arp 信息 10.244.2.0 ether 3e:0a:47:72:56:37 CM flannel.1、fdb 信息 3e:0a:47:72:56:37 dev flannel.1 dst 10.1.5.11 self permanent
数据封装完成后，会送往 net0 接口，并送往 gw0 主机。
gw0 主机接受到数据包后，发目的地址为 10.1.5.11，会查看路由表，送往 eth1 接口。路由: 10.1.5.0/24 dev eth1 proto kernel scope link src 10.1.5.1
通过 gw0 主机 eth1 网卡重新封装数据包后，最终会把数据包送到 clab-flannel-vxlan-directrouting-worker 主机
对端 clab-flannel-vxlan-directrouting-worker 主机接受到数据包后，发现这个是本机数据包信息，在解封装数据包过程中发现这是一个送往 UDP 8472 接口的 vxlan 数据包，将数据包交给监听 UDP 8472 端口的应用程序或内核模块处理。
解封装后发现内部的数据包，目的地址为 10.244.2.2 ，通过查看本机路由表，送往 cni0 网卡。路由: 10.244.1.0/24 dev cni0 proto kernel scope link src 10.244.1.1
通过 cni0 网卡 brctl showmacs cni0 mac 信息，最终会把数据包送到 cni-x2rxr 主机