k8s实践10:docker网桥和pod连接网络结构
1.1
使用到的组件
10年的云岩网站建设经验,针对设计、前端、开发、售后、文案、推广等六对一服务,响应快,48小时及时工作处理。成都营销网站建设的优势是能够根据用户设备显示端的尺寸不同,自动调整云岩建站的显示方式,使网站能够适用不同显示终端,在浏览器中调整网站的宽度,无论在任何一种浏览器上浏览网站,都能展现优雅布局与设计,从而大程度地提升浏览体验。创新互联建站从事“云岩网站设计”,“云岩网站推广”以来,每个客户项目都认真落实执行。
network namespace
bridge
veth pair
注意事项:所有命令都是在集群master节点操作测试.
1.2
network namespace
执行命令ip netns会显示创建的network namespace
如果显示空白,请做link,命令见下:
[root@k8s-master2 ~]# ln -s /var/run/docker/netns/ /var/run/netns
[root@k8s-master1 ~]# ip netns
default
[root@k8s-master1 ~]#
[root@k8s-master2 ~]# ip netns
netns
290d0c6ad2ab (id: 3)
49bfcc08baa8 (id: 2)
aeb6eaef384f (id: 1)
8d103e06202b (id: 0)
default
[root@k8s-master2 ~]#
[root@k8s-master3 ~]# ip netns
789ea14c2216 (id: 4)
a4dbfec964e5 (id: 3)
4bdb1409b6e3 (id: 2)
daea28f1e8b2 (id: 1)
0566deb27814 (id: 0)
default
[root@k8s-master3 ~]#
可以看到每个节点显示的内容不同,默认只有一个default.
1.3
为什么每个节点显示的不同?
以master3做示例,见下
[root@k8s-master3 ~]# ip a |grep veth
8: veth3a37cd9@if7: mtu 1450 qdisc noqueue master docker0 state UP
10: veth37e9e7e@if9: mtu 1450 qdisc noqueue master docker0 state UP
12: veth4ed5bcb@if11: mtu 1450 qdisc noqueue master docker0 state UP
14: vethc4d2905@if13: mtu 1450 qdisc noqueue master docker0 state UP
16: vethd67ea28@if15: mtu 1450 qdisc noqueue master docker0 state UP
[root@k8s-master3 ~]#
[root@k8s-master3 ~]# ip netns
789ea14c2216 (id: 4)
a4dbfec964e5 (id: 3)
4bdb1409b6e3 (id: 2)
daea28f1e8b2 (id: 1)
0566deb27814 (id: 0)
default
[root@k8s-master3 ~]#
注意,一个netns对应了一个虚拟的网卡.怎么对应的呢?
示例,见下:
[root@k8s-master3 ~]# ip netns exec 0566deb27814 ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
7: eth0@if8: mtu 1450 qdisc noqueue state UP
link/ether 02:42:ac:1e:5b:02 brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 172.30.91.2/24 scope global eth0
valid_lft forever preferred_lft forever
inet6 fe80::42:acff:fe1e:5b02/64 scope link
valid_lft forever preferred_lft forever
请看这条:
eth0@if8: mtu 1450 qdisc noqueue state UP
对应的是:
8: veth3a37cd9@if7: mtu 1450 qdisc noqueue master docker0 state UP
这是一个veth pair对.
可用命令
[root@k8s-master3 ~]# ethtool -i veth3a37cd9
driver: veth
version: 1.0
firmware-version:
expansion-rom-version:
bus-info:
supports-statistics: yes
supports-test: no
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: no
[root@k8s-master3 ~]# ethtool -S veth3a37cd9
NIC statistics:
peer_ifindex: 7
[root@k8s-master3 ~]#
回到开始的问题,为什么每个节点显示的不同呢?
network namespace是怎么来的呢?
先敲命令做显示对比.
pod
[root@k8s-master3 ~]# kubectl get pod -n kube-system -o wide
NAME READY STATUS RESTARTS AGE IP NODE
coreDNS-779ffd89bd-5znxf 1/1 Running 2 18h 172.30.91.3 k8s-master3
kubernetes-dashboard-65c76f6c97-4kg2r 1/1 Running 3 18h 172.30.91.5 k8s-master3
tiller-deploy-58d57fd669-w8wsz 1/1 Running 3 18h 172.30.91.2 k8s-master3
traefik-ingress-controller-77b549b5d9-4xd58 1/1 Running 4 4d 192.168.32.130 k8s-master3
traefik-ingress-controller-77b549b5d9-dj65f 1/1 Running 10 34d 192.168.32.129 k8s-master2
traefik-ingress-controller-77b549b5d9-hpt4z 1/1 Running 11 34d 192.168.32.128 k8s-master1
[root@k8s-master3 ~]#
pod的ip
[root@k8s-master3 ~]# kubectl exec -it tiller-deploy-58d57fd669-w8wsz /bin/sh -n kube-system
~ $ ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
7: eth0@if8: mtu 1450 qdisc noqueue state UP
link/ether 02:42:ac:1e:5b:02 brd ff:ff:ff:ff:ff:ff
inet 172.30.91.2/24 scope global eth0
valid_lft forever preferred_lft forever
inet6 fe80::42:acff:fe1e:5b02/64 scope link
valid_lft forever preferred_lft forever
eth0@if8,看看这个network namespace的ip情况
[root@k8s-master3 ~]# ip netns exec 0566deb27814 ip a
1: lo: mtu 65536 qdisc noqueue state UNKNOWN qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
7: eth0@if8: mtu 1450 qdisc noqueue state UP
link/ether 02:42:ac:1e:5b:02 brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 172.30.91.2/24 scope global eth0
valid_lft forever preferred_lft forever
inet6 fe80::42:acff:fe1e:5b02/64 scope link
valid_lft forever preferred_lft forever
一个pod对应了一个network namespace,pod使用的网络连接口就是network namespace的网络口.
1.4
bridge
执行命令,见下:
[root@k8s-master3 ~]# brctl show
bridge name bridge id STP enabled interfaces
docker0 8000.0242c5980278 no veth37e9e7e
veth3a37cd9
veth4ed5bcb
vethc4d2905
vethd67ea28
[root@k8s-master3 ~]# ip a |grep veth
8: veth3a37cd9@if7: mtu 1450 qdisc noqueue master docker0 state UP
10: veth37e9e7e@if9: mtu 1450 qdisc noqueue master docker0 state UP
12: veth4ed5bcb@if11: mtu 1450 qdisc noqueue master docker0 state UP
14: vethc4d2905@if13: mtu 1450 qdisc noqueue master docker0 state UP
16: vethd67ea28@if15: mtu 1450 qdisc noqueue master docker0 state UP
结论:
这些虚拟网卡每个对应的eth pair一端(另一端在namaspace里其实也就是在pod里,参考上面的分析),同时这些虚拟网卡桥接到了网桥docker上.
1.5
网桥docker和pod的逻辑拓扑关系图
拓扑见下:
新闻名称:k8s实践10:docker网桥和pod连接网络结构
新闻来源:http://hbruida.cn/article/jsoeig.html