100、神器的 routing mesh （Swarm07）
上一节我们提到了 swarm 的 routing mesh 。当外部访问任意节点的8080端口时，swarm 内部的 load balance 会将请求转发给web_server 其中的一个副本。大概如下图所示：
所以，无论访问哪个节点，及时该节点上没有运行Service的副本，最终都能访问到Service。
另外，我们还可以配置一个外部的 load balance ，将请求路由到 swarm Service 。比如配置HAProxy，将请求分发到各个节点的8080端口。
如下图所示，一共是两层 load balance。
ingress 网络
当我们应用 --publis-add 8080:80 时，swarm 会重新配置service，我们会看到容器发生了以下变化
root@host03:~# docker service create --name web_server --replicas 2 tomcat
v2p0yuexiws1g1qs2xj5hmqte
overall progress: 2 out of 2 tasks
1/2: running [==================================================>]
2/2: running [==================================================>]
verify: Service converged
root@host03:~# docker service ps web_server
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
be17grtfyj33 web_server.1 tomcat:latest host02 Running Shutdown 11 seconds ago
wsfub26hj5tq web_server.2 tomcat:latest host01 Running Shutdown 14 seconds ago
root@host01:~# docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
1bb657e47d1f tomcat:latest "catalina.sh run" 16 seconds ago Up 14 seconds 8080/tcp web_server.2.wsfub26hj5tqvufrtvir6nmip
root@host02:~# docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
1183f0b73cba tomcat:latest "catalina.sh run" 19 seconds ago Up 17 seconds 8080/tcp web_server.1.be17grtfyj33nbhyyrhe8dh4c
root@host03:~# docker service update --publish-add 8080:8080 web_server
web_server
overall progress: 2 out of 2 tasks
1/2: running [==================================================>]
2/2: running [==================================================>]
verify: Service converged
root@host03:~# docker service ps web_server
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
0y0wh18zbw0l web_server.1 tomcat:latest host02 Running 10 seconds ago
be17grtfyj33 \_ web_server.1 tomcat:latest host02 Shutdown Shutdown 11 seconds ago
lvhvtel79hr8 web_server.2 tomcat:latest host01 Running Running 13 seconds ago
wsfub26hj5tq \_ web_server.2 tomcat:latest host01 Shutdown Shutdown 14 seconds ago
root@host01:~# docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
2fa7f422dfb6 tomcat:latest "catalina.sh run" 24 seconds ago Up 21 seconds 8080/tcp web_server.2.lvhvtel79hr8yw8fzeoaib0u9
1bb657e47d1f tomcat:latest "catalina.sh run" 2 minutes ago Exited (143) 22 seconds ago web_server.2.wsfub26hj5tqvufrtvir6nmip
root@host02:~# docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
b1ba99790519 tomcat:latest "catalina.sh run" 23 seconds ago Up 20 seconds 8080/tcp web_server.1.0y0wh18zbw0lhpks3zbrqdprc
1183f0b73cba tomcat:latest "catalina.sh run" 2 minutes ago Exited (143) 21 seconds ago web_server.1.be17grtfyj33nbhyyrhe8dh4c
root@host01:~# docker exec web_server.2.lvhvtel79hr8yw8fzeoaib0u9 ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default 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
1141: eth0@if1142: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default
link/ether 02:42:0a:ff:00:0c brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 10.255.0.12/16 brd 10.255.255.255 scope global eth0
valid_lft forever preferred_lft forever
1143: eth1@if1144: <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 1
inet 172.18.0.3/16 brd 172.18.255.255 scope global eth1
valid_lft forever preferred_lft forever
root@host02:~# docker exec web_server.1.0y0wh18zbw0lhpks3zbrqdprc ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default 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
1237: eth0@if1238: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default
link/ether 02:42:0a:ff:00:0d brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 10.255.0.13/16 brd 10.255.255.255 scope global eth0
valid_lft forever preferred_lft forever
1239: eth1@if1240: <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 1
inet 172.18.0.3/16 brd 172.18.255.255 scope global eth1
valid_lft forever preferred_lft forever
root@host01:~# docker network ls
NETWORK ID NAME DRIVER SCOPE
d4b0862f06c8 bridge bridge local
d883d53943b2 docker_gwbridge bridge local
358643b57976 host host local
sngp88bsqode ingress overlay swarm
f5d2888b2321 none null local
root@host02:~# docker network ls
NETWORK ID NAME DRIVER SCOPE
3fd2a4a2ffc0 bridge bridge local
d8c36dcdd92a docker_gwbridge bridge local
3a8925305a80 host host local
sngp88bsqode ingress overlay swarm
ce845103f54d none null local
之前所有的副本都被shutdown了，然后启动了新的副本，在新的副本里面可以看到两块网卡
eth0 连接 overlay 类型的网络，名字是 ingress，其作用是让运行在不同主机上的容器可以互相通信。
eth1 连接的是一个 bridge 类型的网络，名字为 docker_gwbridge，其作用是让容器能都访问到外网
ingress 网络是swarm创建时 Docker 自动为我们创建的，swarm中每个 node 都能使用ingress 。
通过 overlay 网络，主机与容器，容器与容器之间可以互相访问。同时，routing mesh 将外部请求路由到不同主机的容器，从而实现了外部网络对service的访问。
