如何使用Calico实现跨主机Docker网络通信_随笔

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 *** 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 *** 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]
└─$ansible etcd -m ping
192.168.26.101 | SUCCESS => {
    "ansible_facts": {
        "discovered_interpreter_python": "/usr/bin/python"
    },
    "changed": false,
    "ping": "pong"
}
192.168.26.102 | SUCCESS => {
    "ansible_facts": {
        "discovered_interpreter_python": "/usr/bin/python"
    },
    "changed": false,
    "ping": "pong"
}
192.168.26.100 | SUCCESS => {
    "ansible_facts": {
        "discovered_interpreter_python": "/usr/bin/python"
    },
    "changed": false,
    "ping": "pong"
}
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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]
└─$ansible etcd -m shell -a "etcdctl member list"
192.168.26.102 | CHANGED | rc=0 >>
6f2038a018db1103, started, etcd-100, http://192.168.26.100:2380, http://192.168.26.100:2379,http://localhost:2379
bd330576bb637f25, started, etcd-101, http://192.168.26.101:2380, http://192.168.26.101:2379,http://localhost:2379
fbd8a96cbf1c004d, started, etcd-102, http://192.168.26.102:2380, http://192.168.26.100:2379,http://localhost:2379
192.168.26.101 | CHANGED | rc=0 >>
6f2038a018db1103, started, etcd-100, http://192.168.26.100:2380, http://192.168.26.100:2379,http://localhost:2379
bd330576bb637f25, started, etcd-101, http://192.168.26.101:2380, http://192.168.26.101:2379,http://localhost:2379
fbd8a96cbf1c004d, started, etcd-102, http://192.168.26.102:2380, http://192.168.26.100:2379,http://localhost:2379
192.168.26.100 | CHANGED | rc=0 >>
6f2038a018db1103, started, etcd-100, http://192.168.26.100:2380, http://192.168.26.100:2379,http://localhost:2379
bd330576bb637f25, started, etcd-101, http://192.168.26.101:2380, http://192.168.26.101:2379,http://localhost:2379
fbd8a96cbf1c004d, started, etcd-102, http://192.168.26.102:2380, http://192.168.26.100:2379,http://localhost:2379
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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.sock
192.168.26.102 | CHANGED | rc=0 >>
ExecStart=/usr/bin/dockerd -H fd:// --containerd=/run/containerd/containerd.sock
192.168.26.101 | 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]
└─$cat calicoctl.j2
apiVersion: v1
kind: calicoApiConfig
metadata:
spec:
  datastoreType: "etcdv2"
  etcdEndpoints: "http://{{inventory_hostname}}:2379"
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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: v1
kind: calicoApiConfig
metadata:
spec:
  datastoreType: "etcdv2"
  etcdEndpoints: "http://192.168.26.100:2379"
192.168.26.102 | CHANGED | rc=0 >>
apiVersion: v1
kind: calicoApiConfig
metadata:
spec:
  datastoreType: "etcdv2"
  etcdEndpoints: "http://192.168.26.102:2379"
192.168.26.101 | CHANGED | rc=0 >>
apiVersion: v1
kind: calicoApiConfig
metadata:
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:v2.6.12
192.168.26.102 | CHANGED | rc=0 >>
Loaded image: quay.io/calico/node:v2.6.12
192.168.26.101 | 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]
└─$ansible etcd -m shell -a "docker images"
192.168.26.102 | CHANGED | rc=0 >>
REPOSITORY            TAG       IMAGE ID       CREATED       SIZE
quay.io/calico/node   v2.6.12   401cc3e56a1a   3 years ago   281MB
192.168.26.100 | CHANGED | rc=0 >>
REPOSITORY            TAG       IMAGE ID       CREATED       SIZE
quay.io/calico/node   v2.6.12   401cc3e56a1a   3 years ago   281MB
192.168.26.101 | CHANGED | rc=0 >>
REPOSITORY            TAG       IMAGE ID       CREATED       SIZE
quay.io/calico/node   v2.6.12   401cc3e56a1a   3 years ago   281MB
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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 status
No 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 status
No 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 status
No 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 *** 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]
└─$ansible etcd -m shell -a "docker network list"
192.168.26.100 | CHANGED | rc=0 >>
NETWORK ID     NAME      DRIVER    SCOPE
caa87ba3dd86   bridge    bridge    local
58121f89bcdd   calnet1   calico    global
1d63e3ad385f   host      host      local
adc94f172d5f   none      null      local
192.168.26.102 | CHANGED | rc=0 >>
NETWORK ID     NAME      DRIVER    SCOPE
cc37d3c66e2f   bridge    bridge    local
58121f89bcdd   calnet1   calico    global
3b138015d4ab   host      host      local
7481614a7084   none      null      local
192.168.26.101 | CHANGED | rc=0 >>
NETWORK ID     NAME      DRIVER    SCOPE
d0cb224ed111   bridge    bridge    local
58121f89bcdd   calnet1   calico    global
106e1c9fb3d3   host      host      local
f983021e2a02   none      null      local
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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:  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
2: ens32:  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 ens32
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe0f:98f1/64 scope link
       valid_lft forever preferred_lft forever
3: docker0:  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 docker0
       valid_lft forever preferred_lft forever
192.168.26.100 | CHANGED | rc=0 >>
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
2: ens32:  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 ens32
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe8c:e81a/64 scope link
       valid_lft forever preferred_lft forever
3: docker0:  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 docker0
       valid_lft forever preferred_lft forever
192.168.26.101 | CHANGED | rc=0 >>
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
2: ens32:  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 ens32
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe3b:6eef/64 scope link
       valid_lft forever preferred_lft forever
3: docker0:  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 docker0
       valid_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 >>
cf2ff4b65e6343fa6e9afba6e75376b97ac47ea59c35f3c492bb7051c15627f0
192.168.26.100 | CHANGED | rc=0 >>
065724c073ded04d6df41d295be3cd5585f8683664fd42a3953dc8067195c58e
192.168.26.102 | CHANGED | rc=0 >>
82e4d6dfde5a6e51f9a4d4f86909678a42e8d1e2d9bfa6edd9cc258b37dfc2db

查看容器节点信息

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ansible etcd -m shell -a "docker ps"
192.168.26.102 | CHANGED | rc=0 >>
ConTAINER ID   IMAGE                         COMMAND         CREATED              STATUS              PORTS     NAMES
82e4d6dfde5a   busybox                       "sh"            about a minute ago   Up about a minute             192.168.26.102
c2d2ab904d6d   quay.io/calico/node:v2.6.12   "start_runit"   2 hours ago          Up 2 hours                    calico-node
192.168.26.100 | CHANGED | rc=0 >>
ConTAINER ID   IMAGE                         COMMAND         CREATED              STATUS              PORTS     NAMES
065724c073de   busybox                       "sh"            about a minute ago   Up about a minute             192.168.26.100
f0b150a924d9   quay.io/calico/node:v2.6.12   "start_runit"   2 hours ago          Up 2 hours                    calico-node
192.168.26.101 | CHANGED | rc=0 >>
ConTAINER ID   IMAGE                         COMMAND         CREATED              STATUS              PORTS     NAMES
cf2ff4b65e63   busybox                       "sh"            about a minute ago   Up about a minute             192.168.26.101
0e4e6f005797   quay.io/calico/node:v2.6.12   "start_runit"   2 hours ago          Up 2 hours                    calico-node
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

查看每个容器的内部网卡和IP

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ansible etcd -m shell -a "docker exec -it {{inventory_hostname}} ip a | grep  cali0 -A 4"
192.168.26.100 | CHANGED | rc=0 >>
4: cali0@if5:  mtu 1500 qdisc noqueue
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff
    inet 192.168.239.128/32 scope global cali0
       valid_lft forever preferred_lft forever
192.168.26.102 | CHANGED | rc=0 >>
4: cali0@if5:  mtu 1500 qdisc noqueue
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff
    inet 192.168.63.64/32 scope global cali0
       valid_lft forever preferred_lft forever
192.168.26.101 | CHANGED | rc=0 >>
4: cali0@if5:  mtu 1500 qdisc noqueue
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff
    inet 192.168.198.0/32 scope global cali0
       valid_lft forever preferred_lft forever
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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 cali0
169.254.1.1 dev cali0 scope link
192.168.26.102 | CHANGED | rc=0 >>
default via 169.254.1.1 dev cali0
169.254.1.1 dev cali0 scope link
192.168.26.100 | CHANGED | rc=0 >>
default via 169.254.1.1 dev cali0
169.254.1.1 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]
└─$ansible etcd -m shell -a "ip a | grep -A 4 cali"
192.168.26.102 | CHANGED | rc=0 >>
5: cali6f956c2ada9@if4:  mtu 1500 qdisc noqueue state UP
    link/ether 6a:65:54:1a:19:e6 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::6865:54ff:fe1a:19e6/64 scope link
       valid_lft forever preferred_lft forever
192.168.26.100 | CHANGED | rc=0 >>
5: cali0b7f49da20a@if4:  mtu 1500 qdisc noqueue state UP
    link/ether 9e:da:0e:cc:b3:7e brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::9cda:eff:fecc:b37e/64 scope link
       valid_lft forever preferred_lft forever
192.168.26.101 | CHANGED | rc=0 >>
5: calib6f7ddae7e3@if4:  mtu 1500 qdisc noqueue state UP
    link/ether 1e:e6:16:ae:f0:91 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::1ce6:16ff:feae:f091/64 scope link
       valid_lft forever preferred_lft forever
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

查看宿主机路由关系

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ansible etcd -m shell -a "ip route "
192.168.26.101 | CHANGED | rc=0 >>
default via 192.168.26.2 dev ens32
169.254.0.0/16 dev ens32 scope link metric 1002
172.17.0.0/16 dev docker0 proto kernel scope link src 172.17.0.1
192.168.26.0/24 dev ens32 proto kernel scope link src 192.168.26.101
192.168.63.64/26 via 192.168.26.102 dev ens32 proto bird
blackhole 192.168.198.0/26 proto bird
192.168.198.1 dev cali2f9e2c68bad scope link
192.168.239.128/26 via 192.168.26.100 dev ens32 proto bird
192.168.26.100 | CHANGED | rc=0 >>
default via 192.168.26.2 dev ens32
169.254.0.0/16 dev ens32 scope link metric 1002
172.17.0.0/16 dev docker0 proto kernel scope link src 172.17.0.1
192.168.26.0/24 dev ens32 proto kernel scope link src 192.168.26.100
192.168.63.64/26 via 192.168.26.102 dev ens32 proto bird
192.168.198.0/26 via 192.168.26.101 dev ens32 proto bird
192.168.239.128 dev cali0b7f49da20a scope link
blackhole 192.168.239.128/26 proto bird
192.168.26.102 | CHANGED | rc=0 >>
default via 192.168.26.2 dev ens32
169.254.0.0/16 dev ens32 scope link metric 1002
172.17.0.0/16 dev docker0 proto kernel scope link src 172.17.0.1
192.168.26.0/24 dev ens32 proto kernel scope link src 192.168.26.102
192.168.63.64 dev cali6f956c2ada9 scope link
blackhole 192.168.63.64/26 proto bird
192.168.198.0/26 via 192.168.26.101 dev ens32 proto bird
192.168.239.128/26 via 192.168.26.100 dev ens32 proto bird
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

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 table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.26.2    0.0.0.0         UG    0      0        0 ens32
169.254.0.0     0.0.0.0         255.255.0.0     U     1002   0        0 ens32
172.17.0.0      0.0.0.0         255.255.0.0     U     0      0        0 docker0
192.168.26.0    0.0.0.0         255.255.255.0   U     0      0        0 ens32
192.168.63.64   192.168.26.102  255.255.255.192 UG    0      0        0 ens32
192.168.198.0   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 cali2f9e2c68bad
192.168.239.128 192.168.26.100  255.255.255.192 UG    0      0        0 ens32
192.168.26.100 | CHANGED | rc=0 >>
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.26.2    0.0.0.0         UG    0      0        0 ens32
169.254.0.0     0.0.0.0         255.255.0.0     U     1002   0        0 ens32
172.17.0.0      0.0.0.0         255.255.0.0     U     0      0        0 docker0
192.168.26.0    0.0.0.0         255.255.255.0   U     0      0        0 ens32
192.168.63.64   192.168.26.102  255.255.255.192 UG    0      0        0 ens32
192.168.198.0   192.168.26.101  255.255.255.192 UG    0      0        0 ens32
192.168.239.128 0.0.0.0         255.255.255.255 UH    0      0        0 cali0b7f49da20a
192.168.239.128 0.0.0.0         255.255.255.192 U     0      0        0 *
192.168.26.102 | CHANGED | rc=0 >>
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.26.2    0.0.0.0         UG    0      0        0 ens32
169.254.0.0     0.0.0.0         255.255.0.0     U     1002   0        0 ens32
172.17.0.0      0.0.0.0         255.255.0.0     U     0      0        0 docker0
192.168.26.0    0.0.0.0         255.255.255.0   U     0      0        0 ens32
192.168.63.64   0.0.0.0         255.255.255.255 UH    0      0        0 cali6f956c2ada9
192.168.63.64   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 ens32
192.168.239.128 192.168.26.100  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
/ # ls
bin   dev   etc   home  proc  root  sys   tmp   usr   var
/ # ip a
1: lo:  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 lo
       valid_lft forever preferred_lft forever
4: cali0@if5:  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 cali0
       valid_lft forever preferred_lft forever
/ # ping 192.168.63.64
PING 192.168.63.64 (192.168.63.64): 56 data bytes
64 bytes from 192.168.63.64: seq=0 ttl=62 time=18.519 ms
64 bytes from 192.168.63.64: seq=1 ttl=62 time=0.950 ms
64 bytes from 192.168.63.64: seq=2 ttl=62 time=1.086 ms
64 bytes from 192.168.63.64: seq=3 ttl=62 time=0.846 ms
64 bytes from 192.168.63.64: seq=4 ttl=62 time=0.840 ms
64 bytes from 192.168.63.64: seq=5 ttl=62 time=1.151 ms
64 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 loss
round-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]
└─$ip a
1: lo:  mtu 65536 qdisc noqueue state UNKNOWN qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: ens32:  mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:0c:29:ad:e3:93 brd ff:ff:ff:ff:ff:ff
    inet 192.168.26.81/24 brd 192.168.26.255 scope global ens32
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fead:e393/64 scope link
       valid_lft forever preferred_lft forever
3: docker0:  mtu 1500 qdisc noqueue state DOWN
    link/ether 02:42:0a:9e:7d:44 brd ff:ff:ff:ff:ff:ff
    inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0
       valid_lft forever preferred_lft forever
4: tunl0@NONE:  mtu 1480 qdisc noqueue state UNKNOWN qlen 1
    link/ipip 0.0.0.0 brd 0.0.0.0
    inet 10.244.88.64/32 scope global tunl0
       valid_lft forever preferred_lft forever
5: cali12cf25006b5@if4:  mtu 1480 qdisc noqueue state UP
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link
       valid_lft forever preferred_lft forever
6: cali5a282a7bbb0@if4:  mtu 1480 qdisc noqueue state UP
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 1
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link
       valid_lft forever preferred_lft forever
7: calicb34164ec79@if4:  mtu 1480 qdisc noqueue state UP
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 2
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link
       valid_lft forever preferred_lft forever
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

欢迎分享，转载请注明来源：内存溢出

原文地址: http://outofmemory.cn/zaji/5693878.html

如何使用Calico实现跨主机Docker网络通信

发表评论

评论列表（0条）