如何使用Calico实现跨主机Docker网络通信
写在前面
- 学习
K8s，学到这里，整理记忆， - 本文用于理解
K8s网络中Calico网路方案的原理优缺点 - 内容包括：
- 常见的容器跨主机通信方案简介
Caliao实现Container跨主机通信简述Calico多个主机通信的实战Demo.没有涉及K8s，通过Docker实现- 博文部分理论来自
《基于Kubernetes的容器云平台实战》，看的PDF，书很不错，感兴趣小伙伴可以支持作者一波。
户外依然大雨滂沱，只是这回彷彿不仅命运一人独自哭泣，不晓得由来，窗外的雨水似乎渗上我心头，有些寒冻，有些缩麻，还有些苦涩。城市万家灯火，橘黄街灯与家户里的温暖流洩，我总觉得这时候的我，最脆弱。 -----《Unser Leben Unser Traum》
跨主机Docker网络通信
常见的跨主机通信方案主要有以下几种:
| 形式 | 描述 |
|---|---|
| Host模式 | 容器直接使用宿主机的网络,这样天生就可以支持跨主机通信。这种方式虽然可以解决跨主机通信问题,但应用场景很有限,容易出现端口冲突,也无法做到隔离网络环境,一个容器崩溃很可能引起整个宿主机的崩溃。 |
| 端口绑定 | 通过绑定容器端口到宿主机端口,跨主机通信时使用“主机IP+端口的方式访问容器中的服务。显然,这种方式仅能支持网络栈的4层及以上的应用,·并且容器与宿主机紧耦合,很难灵活地处理问题,可扩展性不佳。 |
| 定义容器网络 | 使用Open vSwitch或Flannel等第三方SDN工具,为容器构建可以跨主机通信的网络环境。 这类方案一般要求各个主机上的Dockero网桥的cidr不同,以避免出现IP冲突的问题,限制容器在宿主机上可获取的IP范围。并且在容器需要对集群外提供服务时,需要比较复杂的配置,对部署实施人员的网络技能要求比较高。 |
容器网络发展到现在,形成了两大阵营:
Docker的CNM;Google,Coreos，Kuberenetes主导的CNI
CNM和CNI是**网络规范或者网络体系**,并不是网络实现因此并不关心容器网络的实现方式( Flannel或者Calico等), CNM和CNI关心的只是网络管理。
| 网络类型 | 描述 |
|---|---|
| CNM (Container Network Model) | CNM的优势在于原生,容器网络和Docker容器,生命周期结合紧密;缺点是被Docker “绑架”。支持CNM网络规范的容器网络实现包括:Docker Swarm overlay, Macvlan & IP networkdrivers, Calico, Contiv, Weave等。 |
| CNI ( Container Network Interface) | CNI的优势是兼容其他容器技术(如rkt)及上层编排系统(Kubernetes&Mesos),而且社区活跃势头迅猛;缺点是非Docker原生。支持CNI网络规范的容器网络实现包括: Kubernetes、 Weave,Macvlan, Calico, Flannel, Contiv.Mesos CNI等。 |
但从 网络实现角度,又可分为:
| 网络实现角度 | 描述 |
|---|---|
| 隧道方案 | 隧道方案在laas层的网络中应用也比较多,它的主要缺点是随着节点规模的增长复杂度会提升,而且出了网络问题后跟踪起来比较麻烦,大规模集群情况下这是需要考虑的一个问题 |
| 路由方案 | 一般是基于3层或者2层实现网络隔离和跨主机容器互通的,出了问题也很容易排查。Calico :基于BGP协议的路由方案,支持很细致的ACL控制,对混合云亲和度比较高。Macvlan:从逻辑和Kernel层来看,是隔离性和性能最优的方案。基于二层隔离,所以需要一层路由器支持,大多数云服务商不支持,所以混合云上比较难以实现。 |
calico通信过程
Calico把每个操作系统的协议栈当作一个路由器,认为所有的容器是连在这个路由器上的网络终端,在路由器之间运行标准的路由协议-BGP,然后让它们自己去学习这个网络拓扑该如何转发。
Calico方案其实是一个纯三层的方案,也就是说让 每台机器的协议栈的三层去确保两个容器、跨主机容器之间的三层连通性。其网络模型如图所示。
| 网络模型 |
|---|
 |
对于控制平面,其
每个Calico节点上会运行两个主要的程序
| 程序 | 描述 |
|---|---|
一个是Felix |
它会监听etcd,并从etcd获取事件,如该节点新增容器或者增加IP地址等。当在这个节点上创建出一个容器,并将其网卡、IP, MAC都设置好后,Felix在内核的路由表里面写一条数据,注明这个IP应该配置到这张网卡。 |
一个标准的路由程序 |
,它会从内核里面获取哪一些IP的路由发生了变化,然后通过标准BGP的路由协议扩散到整个其他宿主机上,通知外界这个IP在这里。 |
由于Calico是一种纯三层(网络层)的实现,因此可以避免与二层方案相关的数据包封装的操作,·中间没有任何的NAT,没有任何的Overlay,所以它的转发效率可能是所有方案中最高的。因为它的包直接走原生TCP/IP的协议栈,它的隔离也因为这个栈而变得好做。因为TCP/IP的协议栈提供了一整套的防火墙规则,所以它可以通过iptables的规则达到比较复杂的隔离逻辑。
Calico实现方案
| 拓扑模式 |
|---|
 |
环境准备
这里我们通过calico来进行跨主机容器网络通信过程演示,
ansible网络测试
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m ping192.168.26.101 | SUCCESS => { "ansible_facts": { "discovered_interpreter_python": "/usr/bin/python" }, "changed": false, "ping": "pong"}192.168.26.102 | SUCCESS => { "ansible_facts": { "discovered_interpreter_python": "/usr/bin/python" }, "changed": false, "ping": "pong"}192.168.26.100 | SUCCESS => { "ansible_facts": { "discovered_interpreter_python": "/usr/bin/python" }, "changed": false, "ping": "pong"}┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$etcd集群测试，这里我们已经搭建好一个etcd集群，etcdctl member list查看集群列表
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "etcdctl member list"192.168.26.102 | CHANGED | rc=0 >>6f2038a018db1103, started, etcd-100, http://192.168.26.100:2380, http://192.168.26.100:2379,http://localhost:2379bd330576bb637f25, started, etcd-101, http://192.168.26.101:2380, http://192.168.26.101:2379,http://localhost:2379fbd8a96cbf1c004d, started, etcd-102, http://192.168.26.102:2380, http://192.168.26.100:2379,http://localhost:2379192.168.26.101 | CHANGED | rc=0 >>6f2038a018db1103, started, etcd-100, http://192.168.26.100:2380, http://192.168.26.100:2379,http://localhost:2379bd330576bb637f25, started, etcd-101, http://192.168.26.101:2380, http://192.168.26.101:2379,http://localhost:2379fbd8a96cbf1c004d, started, etcd-102, http://192.168.26.102:2380, http://192.168.26.100:2379,http://localhost:2379192.168.26.100 | CHANGED | rc=0 >>6f2038a018db1103, started, etcd-100, http://192.168.26.100:2380, http://192.168.26.100:2379,http://localhost:2379bd330576bb637f25, started, etcd-101, http://192.168.26.101:2380, http://192.168.26.101:2379,http://localhost:2379fbd8a96cbf1c004d, started, etcd-102, http://192.168.26.102:2380, http://192.168.26.100:2379,http://localhost:2379┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$docker安装启动,修改数据存储位置
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "yum -y install docker-ce"┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "systemctl enable docker --now"┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "systemctl status docker"192.168.26.100 | CHANGED | rc=0 >>● docker.service - Docker Application Container Engine Loaded: loaded (/usr/lib/systemd/system/docker.service; enabled; vendor preset: disabled) Active: active (running) since Sat 2022-01-01 20:27:17 CST; 10min ago Docs: https://docs.docker.com ...修改docker启动参数:数据存储位置--cluster-store=
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "cat /usr/lib/systemd/system/docker.service | grep containerd.sock"192.168.26.100 | CHANGED | rc=0 >>ExecStart=/usr/bin/dockerd -H fd:// --containerd=/run/containerd/containerd.sock192.168.26.102 | CHANGED | rc=0 >>ExecStart=/usr/bin/dockerd -H fd:// --containerd=/run/containerd/containerd.sock192.168.26.101 | CHANGED | rc=0 >>ExecStart=/usr/bin/dockerd -H fd:// --containerd=/run/containerd/containerd.sock这里我们直接使用SED来修改
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible 192.168.26.100 -m shell -a "sed -i 's#containerd\.sock#containerd.sock --cluster-store=etcd://192.168.26.100:2379#' /usr/lib/systemd/system/docker.service "192.168.26.100 | CHANGED | rc=0 >>┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible 192.168.26.101 -m shell -a "sed -i 's#containerd\.sock#containerd.sock --cluster-store=etcd://192.168.26.101:2379#' /usr/lib/systemd/system/docker.service "192.168.26.101 | CHANGED | rc=0 >>┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible 192.168.26.102 -m shell -a "sed -i 's#containerd\.sock#containerd.sock --cluster-store=etcd://192.168.26.102:2379#' /usr/lib/systemd/system/docker.service "192.168.26.102 | CHANGED | rc=0 >>┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$刷新Service文件，重启docker
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "systemctl daemon-reload; systemctl restart docker"192.168.26.100 | CHANGED | rc=0 >>192.168.26.102 | CHANGED | rc=0 >>192.168.26.101 | CHANGED | rc=0 >>┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "systemctl status docker"搭建Calico网络
然后我们需要创建calico配置文件,这里我们通过ansilbe 的方式
使用file模块新建文件夹mkdir /etc/calico
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m file -a "path=/etc/calico/ state=directory force=yes"使用template模块创建配置文件 新建模板，这里使用到j2模板，魔法变量
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$cat calicoctl.j2apiVersion: v1kind: calicoApiConfigmetadata:spec: datastoreType: "etcdv2" etcdEndpoints: "http://{{inventory_hostname}}:2379"┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$calico集群创建配置文件
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m template -a "src=calicoctl.j2 dest=/etc/calico/calicoctl.cfg force=yes"核对创建的配置文件
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "cat /etc/calico/calicoctl.cfg"192.168.26.100 | CHANGED | rc=0 >>apiVersion: v1kind: calicoApiConfigmetadata:spec: datastoreType: "etcdv2" etcdEndpoints: "http://192.168.26.100:2379"192.168.26.102 | CHANGED | rc=0 >>apiVersion: v1kind: calicoApiConfigmetadata:spec: datastoreType: "etcdv2" etcdEndpoints: "http://192.168.26.102:2379"192.168.26.101 | CHANGED | rc=0 >>apiVersion: v1kind: calicoApiConfigmetadata:spec: datastoreType: "etcdv2" etcdEndpoints: "http://192.168.26.101:2379"Calico相关镜像导入
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m copy -a "src=/root/calico-node-v2.tar dest=/root/"┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker load -i /root/calico-node-v2.tar"192.168.26.100 | CHANGED | rc=0 >>Loaded image: quay.io/calico/node:v2.6.12192.168.26.102 | CHANGED | rc=0 >>Loaded image: quay.io/calico/node:v2.6.12192.168.26.101 | CHANGED | rc=0 >>Loaded image: quay.io/calico/node:v2.6.12┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$镜像查看
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker images"192.168.26.102 | CHANGED | rc=0 >>REPOSITORY TAGIMAGE IDCREATEDSIZEquay.io/calico/node v2.6.12 401cc3e56a1a 3 years ago 281MB192.168.26.100 | CHANGED | rc=0 >>REPOSITORY TAGIMAGE IDCREATEDSIZEquay.io/calico/node v2.6.12 401cc3e56a1a 3 years ago 281MB192.168.26.101 | CHANGED | rc=0 >>REPOSITORY TAGIMAGE IDCREATEDSIZEquay.io/calico/node v2.6.12 401cc3e56a1a 3 years ago 281MB┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$calicoctl 工具导入
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m copy -a "src=/root/calicoctl dest=/bin/ mode=+x"开始建立 calico node 信息：每个主机上都部署了Calico/Node作为虚拟路由器
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "calicoctl node run --node-image=quay.io/calico/node:v2.6.12 -c /etc/calico/calicoctl.cfg"| 查看node状态，通过Calico将宿主机组织成任意的拓扑集群 |
|---|
 |
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "calicoctl node status"192.168.26.102 | CHANGED | rc=0 >>Calico process is running.IPv4 BGP status+----------------+-------------------+-------+----------+-------------+| PEER ADDRESS | PEER TYPE | STATE | SINCE | INFO |+----------------+-------------------+-------+----------+-------------+| 192.168.26.100 | node-to-node mesh | up | 14:46:35 | Established || 192.168.26.101 | node-to-node mesh | up | 14:46:34 | Established |+----------------+-------------------+-------+----------+-------------+IPv6 BGP statusNo IPv6 peers found.192.168.26.101 | CHANGED | rc=0 >>Calico process is running.IPv4 BGP status+----------------+-------------------+-------+----------+-------------+| PEER ADDRESS | PEER TYPE | STATE | SINCE | INFO |+----------------+-------------------+-------+----------+-------------+| 192.168.26.100 | node-to-node mesh | up | 14:46:31 | Established || 192.168.26.102 | node-to-node mesh | up | 14:46:34 | Established |+----------------+-------------------+-------+----------+-------------+IPv6 BGP statusNo IPv6 peers found.192.168.26.100 | CHANGED | rc=0 >>Calico process is running.IPv4 BGP status+----------------+-------------------+-------+----------+-------------+| PEER ADDRESS | PEER TYPE | STATE | SINCE | INFO |+----------------+-------------------+-------+----------+-------------+| 192.168.26.101 | node-to-node mesh | up | 14:46:31 | Established || 192.168.26.102 | node-to-node mesh | up | 14:46:35 | Established |+----------------+-------------------+-------+----------+-------------+IPv6 BGP statusNo IPv6 peers found.┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$当集群中的容器需要与外界通信时,就可以通过BGP协议将网关物理路由器加入到集群中,使外界可以直接访问容器IP,而不需要做任何NAT之类的复杂操作。
通过Calico网络实现跨主机通信
在某一个Node上创建一个docker内部calico网络
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible 192.168.26.100 -m shell -a "docker network create --driver calico --ipam-driver calico-ipam calnet1"192.168.26.100 | CHANGED | rc=0 >>58121f89bcddec441770aa207ef662d09e4413625b0827ce4d8f601fb10650d0会发现这个内网网络变成的一个全局的网络，在所有节点可见，58121f89bcdd
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker network list"192.168.26.100 | CHANGED | rc=0 >>NETWORK ID NAME DRIVER SCOPEcaa87ba3dd86 bridge bridge local58121f89bcdd calnet1 calico global1d63e3ad385f host host localadc94f172d5f none null local192.168.26.102 | CHANGED | rc=0 >>NETWORK ID NAME DRIVER SCOPEcc37d3c66e2f bridge bridge local58121f89bcdd calnet1 calico global3b138015d4ab host host local7481614a7084 none null local192.168.26.101 | CHANGED | rc=0 >>NETWORK ID NAME DRIVER SCOPEd0cb224ed111 bridge bridge local58121f89bcdd calnet1 calico global106e1c9fb3d3 host host localf983021e2a02 none null local┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$查看节点中的网卡信息,这个时候没有容器运行，所以没有caliao网卡
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "ip a"192.168.26.102 | CHANGED | rc=0 >>1: lo: <LOOPBACK,UP,LOWER_UP> 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 lovalid_lft forever preferred_lft forever inet6 ::1/128 scope hostvalid_lft forever preferred_lft forever2: ens32: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 link/ether 00:0c:29:0f:98:f1 brd ff:ff:ff:ff:ff:ff inet 192.168.26.102/24 brd 192.168.26.255 scope global ens32valid_lft forever preferred_lft forever inet6 fe80::20c:29ff:fe0f:98f1/64 scope linkvalid_lft forever preferred_lft forever3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN link/ether 02:42:c3:28:19:78 brd ff:ff:ff:ff:ff:ff inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0valid_lft forever preferred_lft forever192.168.26.100 | CHANGED | rc=0 >>1: lo: <LOOPBACK,UP,LOWER_UP> 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 lovalid_lft forever preferred_lft forever inet6 ::1/128 scope hostvalid_lft forever preferred_lft forever2: ens32: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 link/ether 00:0c:29:8c:e8:1a brd ff:ff:ff:ff:ff:ff inet 192.168.26.100/24 brd 192.168.26.255 scope global ens32valid_lft forever preferred_lft forever inet6 fe80::20c:29ff:fe8c:e81a/64 scope linkvalid_lft forever preferred_lft forever3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN link/ether 02:42:f7:1a:2e:30 brd ff:ff:ff:ff:ff:ff inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0valid_lft forever preferred_lft forever192.168.26.101 | CHANGED | rc=0 >>1: lo: <LOOPBACK,UP,LOWER_UP> 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 lovalid_lft forever preferred_lft forever inet6 ::1/128 scope hostvalid_lft forever preferred_lft forever2: ens32: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 link/ether 00:0c:29:3b:6e:ef brd ff:ff:ff:ff:ff:ff inet 192.168.26.101/24 brd 192.168.26.255 scope global ens32valid_lft forever preferred_lft forever inet6 fe80::20c:29ff:fe3b:6eef/64 scope linkvalid_lft forever preferred_lft forever3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN link/ether 02:42:70:a7:4e:7e brd ff:ff:ff:ff:ff:ff inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0valid_lft forever preferred_lft forever┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$每个节点运行一个容器
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker run --name {{inventory_hostname}} -itd --net=calnet1 --restart=always busybox "192.168.26.101 | CHANGED | rc=0 >>cf2ff4b65e6343fa6e9afba6e75376b97ac47ea59c35f3c492bb7051c15627f0192.168.26.100 | CHANGED | rc=0 >>065724c073ded04d6df41d295be3cd5585f8683664fd42a3953dc8067195c58e192.168.26.102 | CHANGED | rc=0 >>82e4d6dfde5a6e51f9a4d4f86909678a42e8d1e2d9bfa6edd9cc258b37dfc2db查看容器节点信息
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker ps"192.168.26.102 | CHANGED | rc=0 >>CONTAINER ID IMAGE COMMAND CREATEDSTATUSPORTS NAMES82e4d6dfde5a busybox "sh" About a minute ago Up About a minute 192.168.26.102c2d2ab904d6d quay.io/calico/node:v2.6.12 "start_runit" 2 hours ago Up 2 hours calico-node192.168.26.100 | CHANGED | rc=0 >>CONTAINER ID IMAGE COMMAND CREATEDSTATUSPORTS NAMES065724c073de busybox "sh" About a minute ago Up About a minute 192.168.26.100f0b150a924d9 quay.io/calico/node:v2.6.12 "start_runit" 2 hours ago Up 2 hours calico-node192.168.26.101 | CHANGED | rc=0 >>CONTAINER ID IMAGE COMMAND CREATEDSTATUSPORTS NAMEScf2ff4b65e63 busybox "sh" About a minute ago Up About a minute 192.168.26.1010e4e6f005797 quay.io/calico/node:v2.6.12 "start_runit" 2 hours ago Up 2 hours calico-node┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$查看每个容器的内部网卡和IP
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker exec -it {{inventory_hostname}} ip a | grep cali0 -A 4"192.168.26.100 | CHANGED | rc=0 >>4: cali0@if5: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff inet 192.168.239.128/32 scope global cali0valid_lft forever preferred_lft forever192.168.26.102 | CHANGED | rc=0 >>4: cali0@if5: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff inet 192.168.63.64/32 scope global cali0valid_lft forever preferred_lft forever192.168.26.101 | CHANGED | rc=0 >>4: cali0@if5: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff inet 192.168.198.0/32 scope global cali0valid_lft forever preferred_lft forever┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$查看容器内的路由关系，即所有的出口都是通过cali0网卡来实现的
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "docker exec -it {{inventory_hostname}} ip route | grep cali0 "192.168.26.101 | CHANGED | rc=0 >>default via 169.254.1.1 dev cali0169.254.1.1 dev cali0 scope link192.168.26.102 | CHANGED | rc=0 >>default via 169.254.1.1 dev cali0169.254.1.1 dev cali0 scope link192.168.26.100 | CHANGED | rc=0 >>default via 169.254.1.1 dev cali0169.254.1.1 dev cali0 scope link每创建一个容器，则会在物理机上创建一张虚拟网卡出来，对应容器中的网卡，从这里可以看到容器里的虚拟网卡 cali0 和物理机的 cali6f956c2ada9 是 veth pair 关系。
关于
veth pair小伙伴可以百度下，这里简单描述，作用很简单，就是要把从一个network namespace发出的数据包转发到另一个namespace。veth设备是成对的，一个是container之中，另一个在container之外(宿主机)，即在真实机器上能看到的。VETH设备总是成对出现，送到一端请求发送的数据总是从另一端以请求接受的形式出现。创建并配置正确后，向其一端输入数据，VETH会改变数据的方向并将其送入内核网络子系统，完成数据的注入，而在另一端则能读到此数据。（Namespace，其中往veth设备上任意一端上RX到的数据，都会在另一端上以TX的方式发送出去）veth工作在L2数据链路层，veth-pair设备在转发数据包过程中并不串改数据包内容。
更多小伙伴可以参考:https://blog.csdn.net/sld880311/article/details/77650937
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "ip a | grep -A 4 cali"192.168.26.102 | CHANGED | rc=0 >>5: cali6f956c2ada9@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP link/ether 6a:65:54:1a:19:e6 brd ff:ff:ff:ff:ff:ff link-netnsid 0 inet6 fe80::6865:54ff:fe1a:19e6/64 scope linkvalid_lft forever preferred_lft forever192.168.26.100 | CHANGED | rc=0 >>5: cali0b7f49da20a@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP link/ether 9e:da:0e:cc:b3:7e brd ff:ff:ff:ff:ff:ff link-netnsid 0 inet6 fe80::9cda:eff:fecc:b37e/64 scope linkvalid_lft forever preferred_lft forever192.168.26.101 | CHANGED | rc=0 >>5: calib6f7ddae7e3@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP link/ether 1e:e6:16:ae:f0:91 brd ff:ff:ff:ff:ff:ff link-netnsid 0 inet6 fe80::1ce6:16ff:feae:f091/64 scope linkvalid_lft forever preferred_lft forever┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$查看宿主机路由关系
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "ip route "192.168.26.101 | CHANGED | rc=0 >>default via 192.168.26.2 dev ens32169.254.0.0/16 dev ens32 scope link metric 1002172.17.0.0/16 dev docker0 proto kernel scope link src 172.17.0.1192.168.26.0/24 dev ens32 proto kernel scope link src 192.168.26.101192.168.63.64/26 via 192.168.26.102 dev ens32 proto birdblackhole 192.168.198.0/26 proto bird192.168.198.1 dev cali2f9e2c68bad scope link192.168.239.128/26 via 192.168.26.100 dev ens32 proto bird192.168.26.100 | CHANGED | rc=0 >>default via 192.168.26.2 dev ens32169.254.0.0/16 dev ens32 scope link metric 1002172.17.0.0/16 dev docker0 proto kernel scope link src 172.17.0.1192.168.26.0/24 dev ens32 proto kernel scope link src 192.168.26.100192.168.63.64/26 via 192.168.26.102 dev ens32 proto bird192.168.198.0/26 via 192.168.26.101 dev ens32 proto bird192.168.239.128 dev cali0b7f49da20a scope linkblackhole 192.168.239.128/26 proto bird192.168.26.102 | CHANGED | rc=0 >>default via 192.168.26.2 dev ens32169.254.0.0/16 dev ens32 scope link metric 1002172.17.0.0/16 dev docker0 proto kernel scope link src 172.17.0.1192.168.26.0/24 dev ens32 proto kernel scope link src 192.168.26.102192.168.63.64 dev cali6f956c2ada9 scope linkblackhole 192.168.63.64/26 proto bird192.168.198.0/26 via 192.168.26.101 dev ens32 proto bird192.168.239.128/26 via 192.168.26.100 dev ens32 proto bird┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$我们那其中一台机器来看：192.168.26.100宿主机来讲
192.168.239.128 dev cali0b7f49da20a scope link
进去：本机到目的地址到 容器IP(192.168.239.128 ) 的数据包都从 cali6f956c2ada9 (新产生的虚拟网卡)走。
192.168.63.64/26 via 192.168.26.102 dev ens32 proto bird
192.168.198.0/26 via 192.168.26.101 dev ens32 proto bird
出来：本机目的地址到 容器IP(192.168.63.64/26) 容器IP(192.168.198.0/26) 网段的数据包都从 ens32 发到 其他的两个宿主机上去。
| 每台主机都知道不同的容器在哪台主机上，所以会动态的设置路由。 |
|---|
 |
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ansible etcd -m shell -a "route -n "192.168.26.101 | CHANGED | rc=0 >>Kernel IP routing tableDestination Gateway Genmask Flags Metric Ref Use Iface0.0.0.0 192.168.26.2 0.0.0.0 UG 0 0 0 ens32169.254.0.0 0.0.0.0 255.255.0.0 U 1002 0 0 ens32172.17.0.0 0.0.0.0 255.255.0.0 U 0 0 0 docker0192.168.26.0 0.0.0.0 255.255.255.0 U 0 0 0 ens32192.168.63.64 192.168.26.102 255.255.255.192 UG 0 0 0 ens32192.168.198.0 0.0.0.0 255.255.255.192 U 0 0 0 *192.168.198.1 0.0.0.0 255.255.255.255 UH 0 0 0 cali2f9e2c68bad192.168.239.128 192.168.26.100 255.255.255.192 UG 0 0 0 ens32192.168.26.100 | CHANGED | rc=0 >>Kernel IP routing tableDestination Gateway Genmask Flags Metric Ref Use Iface0.0.0.0 192.168.26.2 0.0.0.0 UG 0 0 0 ens32169.254.0.0 0.0.0.0 255.255.0.0 U 1002 0 0 ens32172.17.0.0 0.0.0.0 255.255.0.0 U 0 0 0 docker0192.168.26.0 0.0.0.0 255.255.255.0 U 0 0 0 ens32192.168.63.64 192.168.26.102 255.255.255.192 UG 0 0 0 ens32192.168.198.0 192.168.26.101 255.255.255.192 UG 0 0 0 ens32192.168.239.128 0.0.0.0 255.255.255.255 UH 0 0 0 cali0b7f49da20a192.168.239.128 0.0.0.0 255.255.255.192 U 0 0 0 *192.168.26.102 | CHANGED | rc=0 >>Kernel IP routing tableDestination Gateway Genmask Flags Metric Ref Use Iface0.0.0.0 192.168.26.2 0.0.0.0 UG 0 0 0 ens32169.254.0.0 0.0.0.0 255.255.0.0 U 1002 0 0 ens32172.17.0.0 0.0.0.0 255.255.0.0 U 0 0 0 docker0192.168.26.0 0.0.0.0 255.255.255.0 U 0 0 0 ens32192.168.63.64 0.0.0.0 255.255.255.255 UH 0 0 0 cali6f956c2ada9192.168.63.64 0.0.0.0 255.255.255.192 U 0 0 0 *192.168.198.0 192.168.26.101 255.255.255.192 UG 0 0 0 ens32192.168.239.128 192.168.26.100 255.255.255.192 UG 0 0 0 ens32┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$简单测试一下：192.168.26.100宿主机上的容器(192.168.239.128)去ping 192.168.63.64(192.168.26.100上的容器)，实现跨主机互通。
┌──[root@vms100.liruilongs.github.io]-[~]└─$ docker exec -it 192.168.26.100 /bin/sh/ # lsbin dev etc home proc root sys tmp usr var/ # ip a1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue 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 lovalid_lft forever preferred_lft forever4: cali0@if5: <BROADCAST,MULTICAST,UP,LOWER_UP,M-DOWN> mtu 1500 qdisc noqueue link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff inet 192.168.239.128/32 scope global cali0valid_lft forever preferred_lft forever/ # ping 192.168.63.64PING 192.168.63.64 (192.168.63.64): 56 data bytes64 bytes from 192.168.63.64: seq=0 ttl=62 time=18.519 ms64 bytes from 192.168.63.64: seq=1 ttl=62 time=0.950 ms64 bytes from 192.168.63.64: seq=2 ttl=62 time=1.086 ms64 bytes from 192.168.63.64: seq=3 ttl=62 time=0.846 ms64 bytes from 192.168.63.64: seq=4 ttl=62 time=0.840 ms64 bytes from 192.168.63.64: seq=5 ttl=62 time=1.151 ms64 bytes from 192.168.63.64: seq=6 ttl=62 time=0.888 ms^C--- 192.168.63.64 ping statistics ---7 packets transmitted, 7 packets received, 0% packet lossround-trip min/avg/max = 0.840/3.468/18.519 ms/ #在K8s集群的中，有一个容器，就会生成一个calico网卡
┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$ip a1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lovalid_lft forever preferred_lft forever inet6 ::1/128 scope hostvalid_lft forever preferred_lft forever2: ens32: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 link/ether 00:0c:29:ad:e3:93 brd ff:ff:ff:ff:ff:ff inet 192.168.26.81/24 brd 192.168.26.255 scope global ens32valid_lft forever preferred_lft forever inet6 fe80::20c:29ff:fead:e393/64 scope linkvalid_lft forever preferred_lft forever3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN link/ether 02:42:0a:9e:7d:44 brd ff:ff:ff:ff:ff:ff inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0valid_lft forever preferred_lft forever4: tunl0@NONE: <NOARP,UP,LOWER_UP> mtu 1480 qdisc noqueue state UNKNOWN qlen 1 link/ipip 0.0.0.0 brd 0.0.0.0 inet 10.244.88.64/32 scope global tunl0valid_lft forever preferred_lft forever5: cali12cf25006b5@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1480 qdisc noqueue state UP link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 0 inet6 fe80::ecee:eeff:feee:eeee/64 scope linkvalid_lft forever preferred_lft forever6: cali5a282a7bbb0@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1480 qdisc noqueue state UP link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 1 inet6 fe80::ecee:eeff:feee:eeee/64 scope linkvalid_lft forever preferred_lft forever7: calicb34164ec79@if4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1480 qdisc noqueue state UP link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 2 inet6 fe80::ecee:eeff:feee:eeee/64 scope linkvalid_lft forever preferred_lft forever┌──[root@vms81.liruilongs.github.io]-[~/ansible]└─$