#云原生征文# Flannel 同节点通信 原创 精华

Flannel 同节点通信

Flannel 同节点通信利用了 veth pair 对的形式，一端连接到 pod，一端连接到宿主机的网卡，同时作为宿主机 cni0 网卡的的接口，使得同节点的通信，直接是二层互通。
和 calico vxlan 不一样的是，pod 对应的网口并没有使用 proxy-arp 的特性，所以就和传统的二层交换，三层路由一样，理解起来更为简单。

集群环境

kubernetes v1.23.5
	master 192.168.0.80
	node1  192.168.0.81
	node2  192.168.0.82

运行 pod

在集群中运行 pod ，查看在 node1 上的 pod
pod1 10.244.1.2
pod2 10.244.1.3

[root@master ~]# kubectl create deployment cni-test --image=burlyluo/nettoolbox --replicas=3

[root@master ~]# kubectl get pod -o wide 
NAME                        READY   STATUS    RESTARTS   AGE   IP           NODE              NOMINATED NODE   READINESS GATES
cni-test-777bbd57c8-5zhjd   1/1     Running   0          18s   10.244.2.6   node2.whale.com   <none>           <none>
cni-test-777bbd57c8-t6xhd   1/1     Running   0          18s   10.244.1.2   node1.whale.com   <none>           <none>
cni-test-777bbd57c8-whwcm   1/1     Running   0          18s   10.244.1.3   node1.whale.com   <none>           <none>

查看 pod1 内部网卡和 node1 对应网卡

[root@master ~]# kubectl exec -it cni-test-777bbd57c8-t6xhd -- bash
bash-5.1# ifconfig 
eth0      Link encap:Ethernet  HWaddr 6A:6D:35:69:37:FE  
          inet addr:10.244.1.2  Bcast:10.244.1.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1450  Metric:1
          RX packets:14 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:1220 (1.1 KiB)  TX bytes:42 (42.0 B)
bash-5.1# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         10.244.1.1      0.0.0.0         UG    0      0        0 eth0
10.244.0.0      10.244.1.1      255.255.0.0     UG    0      0        0 eth0
10.244.1.0      0.0.0.0         255.255.255.0   U     0      0        0 eth0
bash-5.1# ethtool -S eth0
NIC statistics:
     peer_ifindex: 6
     rx_queue_0_xdp_packets: 0
     rx_queue_0_xdp_bytes: 0
     rx_queue_0_xdp_drops: 0

# node1 对应 pod1 网卡
[root@node1 ~]# ip link show | grep ^6
6: veth9903a3d4@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master cni0 state UP mode DEFAULT group default 
[root@node1 ~]# ip -d link show veth9903a3d4
6: veth9903a3d4@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master cni0 state UP mode DEFAULT group default 
    link/ether 2a:5e:4a:15:33:1c brd ff:ff:ff:ff:ff:ff link-netnsid 0 promiscuity 1 
    veth 
    bridge_slave state forwarding priority 32 cost 2 hairpin on guard off root_block off fastleave off learning on flood on port_id 0x8001 port_no 0x1 designated_port 32769 designated_cost 0 designated_bridge 8000.9a:50:49:2d:65:48 designated_root 8000.9a:50:49:2d:65:48 hold_timer    0.00 message_age_timer    0.00 forward_delay_timer    0.00 topology_change_ack 0 config_pending 0 proxy_arp off proxy_arp_wifi off mcast_router 1 mcast_fast_leave off mcast_flood on addrgenmode eui64 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535 

查看 pod2 内部网卡和 node1 对应网卡

[root@master ~]# kubectl exec -it cni-test-777bbd57c8-whwcm -- bash
bash-5.1# ifconfig 
eth0      Link encap:Ethernet  HWaddr CA:BB:AF:BD:E5:3A  
          inet addr:10.244.1.3  Bcast:10.244.1.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1450  Metric:1
          RX packets:13 errors:0 dropped:0 overruns:0 frame:0
          TX packets:1 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:1178 (1.1 KiB)  TX bytes:42 (42.0 B)
bash-5.1# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         10.244.1.1      0.0.0.0         UG    0      0        0 eth0
10.244.0.0      10.244.1.1      255.255.0.0     UG    0      0        0 eth0
10.244.1.0      0.0.0.0         255.255.255.0   U     0      0        0 eth0
bash-5.1# ethtool -S eth0
NIC statistics:
     peer_ifindex: 7
     rx_queue_0_xdp_packets: 0
     rx_queue_0_xdp_bytes: 0
     rx_queue_0_xdp_drops: 0

# node1 对应 pod2 网卡
[root@node1 ~]# ip -d link show vetha61ebf2f
7: vetha61ebf2f@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master cni0 state UP mode DEFAULT group default 
    link/ether b6:c0:6c:fd:e6:e0 brd ff:ff:ff:ff:ff:ff link-netnsid 1 promiscuity 1 
    veth 
    bridge_slave state forwarding priority 32 cost 2 hairpin on guard off root_block off fastleave off learning on flood on port_id 0x8002 port_no 0x2 designated_port 32770 designated_cost 0 designated_bridge 8000.9a:50:49:2d:65:48 designated_root 8000.9a:50:49:2d:65:48 hold_timer    0.00 message_age_timer    0.00 forward_delay_timer    0.00 topology_change_ack 0 config_pending 0 proxy_arp off proxy_arp_wifi off mcast_router 1 mcast_fast_leave off mcast_flood on addrgenmode eui64 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535 

查看 node1 网卡

我们可以见到 pod 对应的 veth 网卡会有 master cni0 选项，这个意思是 cni0 的桥接网卡的接口。

通过查看 cni0 网卡接口信息，确定 pod 对端的 veth pair 网卡是在 cni0 上的接口

通过查看 node1 的路由，我们可以看到 10.244.1.0 指向了 cni0 网卡，所以可以确认，同节点的通信经过的都是 cni0 网桥，而两端的接口分别指向了两个 pod 的对端的 MAC 地址。

[root@node1 ~]# route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
10.244.1.0      0.0.0.0         255.255.255.0   U     0      0        0 cni0

[root@node1 ~]# brctl show cni0
bridge name	bridge id		STP enabled	interfaces
cni0		8000.9a50492d6548	no		veth9903a3d4
										vetha61ebf2f
[root@node1 ~]# brctl showmacs cni0
port no	mac addr		is local?	ageing timer
  1	2a:5e:4a:15:33:1c	yes		   0.00
  1	2a:5e:4a:15:33:1c	yes		   0.00
  2	b6:c0:6c:fd:e6:e0	yes		   0.00
  2	b6:c0:6c:fd:e6:e0	yes		   0.00

同节点通信抓包演示

针对同节点通信，我们通过抓包演示
针对 pod eth0 网卡以及对应 veth 网卡，还有 cni0 进行抓包

kubectl exec -it cni-test-777bbd57c8-t6xhd -- ping -c 1 10.244.1.3

pod1.cap

kubectl exec -it cni-test-777bbd57c8-t6xhd -- tcpdump -pne -i eth0 -w pod1.cap

pod1-veth.cap

tcpdump -pne -i veth9903a3d4 -w pod1-veth.cap

cni0.cap

tcpdump -pne -i cni0 -w cni0.cap

pod2-veth.cap

tcpdump -pne -i vetha61ebf2f -w pod2-veth.cap

pod2.cap

kubectl exec -it cni-test-777bbd57c8-whwcm -- tcpdump -pne -i eth0 -w pod2.cap