这一篇主要是对 Docker Swarm 的完善，增加基于 overlay 组网通信，以便 Docker 容器可以跨主机访问。
不同主机间的容器之间通信方式，大概有三种：
- 使用端口映射：直接把容器的服务端口映射到主机上，主机直接通过映射出来的端口通信。
- 把容器放到主机所在的网段：修改 docker 的 ip 分配网段和主机一致，还要修改主机的网络结构。
- 第三方项目：flannel，weave 或者 pipework 等，这些方案一般都是通过 SDN 搭建 overlay 网络达到容器通信的。
在使用 overlay 组网通信之前，我们先安装 Docker，以及 Docker Machine（Linux 下）：
$ sudo curl -L https://github.com/docker/machine/releases/download/v0.13.0/docker-machine-`uname -s`-`uname -m` > /usr/local/bin/docker-machine
$ sudo chmod +x /usr/local/bin/docker-machine
使用脚本一键安装（Docker 镜像加速地址可以更换）：
#!/bin/bash
set -e
create_kv() {
echo Creating kvstore machine.
docker-machine create -d virtualbox
--engine-opt="registry-mirror=https://kvo9moak.mirror.aliyuncs.com"
kvstore
docker $(docker-machine config kvstore) run -d
-p "8500:8500"
progrium/consul --server -bootstrap-expect 1
}
create_master() {
echo Creating cluster master
kvip=$(docker-machine ip kvstore)
docker-machine create -d virtualbox
--swarm --swarm-master
--swarm-discovery="consul://${kvip}:8500"
--engine-opt="cluster-store=consul://${kvip}:8500"
--engine-opt="cluster-advertise=eth1:2376"
--engine-opt="registry-mirror=https://kvo9moak.mirror.aliyuncs.com"
swarm-manager
}
create_nodes(){
kvip=$(docker-machine ip kvstore)
echo Creating cluster nodes
for i in 1 2; do
docker-machine create -d virtualbox
--swarm
--swarm-discovery="consul://${kvip}:8500"
--engine-opt="cluster-store=consul://${kvip}:8500"
--engine-opt="cluster-advertise=eth1:2376"
--engine-opt="registry-mirror=https://kvo9moak.mirror.aliyuncs.com"
swarm-node${i}
done
}
teardown(){
docker-machine rm kvstore -y
docker-machine rm -y swarm-manager
for i in 1 2; do
docker-machine rm -y swarm-node${i}
done
}
case $1 in
up)
create_kv
create_master
create_nodes
;;
down)
teardown
;;
*)
echo "Unknow command..."
exit 1
;;
esac
运行./cluster.sh up，就能自动生成四台主机：
- 一台 kvstore 运行 consul 服务。
- 一台 swarm master 机器，运行 swarm manager 服务。
- 两台 swarm node 机器，都是运行了 swarm node 服务和 docker daemon 服务。
查看四台主机的具体信息：
$ docker-machine ls
NAME ACTIVE DRIVER STATE URL SWARM DOCKER ERRORS
kvstore - virtualbox Running tcp://192.168.99.100:2376 v18.03.1-ce
swarm-manager * (swarm) virtualbox Running tcp://192.168.99.101:2376 swarm-manager (master) v18.03.1-ce
swarm-node1 - virtualbox Running tcp://192.168.99.102:2376 swarm-manager v18.03.1-ce
swarm-node2 - virtualbox Running tcp://192.168.99.103:2376 swarm-manager v18.03.1-ce
接下来验证集群是否正确安装？在主机上运行下面命令（主机，不是 Docker 主机）：
$ eval $(docker-machine env --swarm swarm-manager)
$ docker info
Containers: 6
Running: 6
Paused: 0
Stopped: 0
Images: 5
Server Version: swarm/1.2.8
Role: primary
Strategy: spread
Filters: health, port, containerslots, dependency, affinity, constraint, whitelist
Nodes: 3
swarm-manager: 192.168.99.101:2376
└ ID: K6WX:ZYFT:UEHA:KM66:BYHD:ROBF:Z5KG:UHNE:U37V:4KX2:S5SV:YSCA|192.168.99.101:2376
└ Status: Healthy
└ Containers: 2 (2 Running, 0 Paused, 0 Stopped)
└ Reserved CPUs: 0 / 1
└ Reserved Memory: 0 B / 1.021 GiB
└ Labels: kernelversion=4.9.93-boot2docker, operatingsystem=Boot2Docker 18.03.1-ce (TCL 8.2.1); HEAD : cb77972 - Thu Apr 26 16:40:36 UTC 2018, ostype=linux, provider=virtualbox, storagedriver=aufs
└ UpdatedAt: 2018-05-08T10:20:39Z
└ ServerVersion: 18.03.1-ce
swarm-node1: 192.168.99.102:2376
└ ID: RPRC:AVBX:7CBJ:HUTI:HI3B:RI6B:QI6O:M2UQ:ZT2I:HZ6J:33BL:HY76|192.168.99.102:2376
└ Status: Healthy
└ Containers: 2 (2 Running, 0 Paused, 0 Stopped)
└ Reserved CPUs: 0 / 1
└ Reserved Memory: 0 B / 1.021 GiB
└ Labels: kernelversion=4.9.93-boot2docker, operatingsystem=Boot2Docker 18.03.1-ce (TCL 8.2.1); HEAD : cb77972 - Thu Apr 26 16:40:36 UTC 2018, ostype=linux, provider=virtualbox, storagedriver=aufs
└ UpdatedAt: 2018-05-08T10:21:09Z
└ ServerVersion: 18.03.1-ce
swarm-node2: 192.168.99.103:2376
└ ID: MKQ2:Y7EO:CKOJ:MGFH:B77S:3EWX:7YJT:2MBQ:CJSN:XENJ:BSKO:RAZP|192.168.99.103:2376
└ Status: Healthy
└ Containers: 2 (2 Running, 0 Paused, 0 Stopped)
└ Reserved CPUs: 0 / 1
└ Reserved Memory: 0 B / 1.021 GiB
└ Labels: kernelversion=4.9.93-boot2docker, operatingsystem=Boot2Docker 18.03.1-ce (TCL 8.2.1); HEAD : cb77972 - Thu Apr 26 16:40:36 UTC 2018, ostype=linux, provider=virtualbox, storagedriver=aufs
└ UpdatedAt: 2018-05-08T10:21:06Z
└ ServerVersion: 18.03.1-ce
Plugins:
Volume:
Network:
Log:
Swarm:
NodeID:
Is Manager: false
Node Address:
Kernel Version: 4.9.93-boot2docker
Operating System: linux
Architecture: amd64
CPUs: 3
Total Memory: 3.063GiB
Name: 85be09a37044
Docker Root Dir:
Debug Mode (client): false
Debug Mode (server): false
Experimental: false
Live Restore Enabled: false
WARNING: No kernel memory limit support
可以看到集群的具体信息。
然后，接下来在主机上创建 overlay 网络，创建命令：
$ docker network create -d overlay net1
d6a8a22298485a044b19fcbb62033ff1b4c3d4bd6a8a2229848
然后我们查看刚创建名称为net1的 overlay 网络，命令：
$ docker network ls
NETWORK ID NAME DRIVER SCOPE
d6a8a2229848 net1 overlay global
9c7f0e793838 swarm-manager/bridge bridge local
93787d9ba7ed swarm-manager/host host local
72fd1e63889e swarm-manager/none null local
c73e00c4c76c swarm-node1/bridge bridge local
95983d8f1ef1 swarm-node1/docker_gwbridge bridge local
a8a569d55cc9 swarm-node1/host host local
e7fa8403b226 swarm-node1/none null local
7f1d219b5c08 swarm-node2/bridge bridge local
e7463ae8c579 swarm-node2/docker_gwbridge bridge local
9a1f0d2bbdf5 swarm-node2/host host local
bea626348d6d swarm-node2/none null local
接下来，我们创建两个容器（主机上执行，使用 Docker Swarm 很方便），测试使用net1网络，是否可以相互访问，创建命令：
$ docker run -d --net=net1 --name=c1 -e constraint:node==swarm-node1 busybox top
dab080b33e76af0e4c71c9365a6e57b2191b7aacd4f715ca11481403eccce12a
$ docker run -d --net=net1 --name=c2 -e constraint:node==swarm-node2 busybox top
583fde42182a7e8f27527d5c55163a32102dba566ebe1f13d1951ac214849c8d
查看刚创建的容器运行情况：
$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
583fde42182a busybox "top" 3 seconds ago Up 2 seconds swarm-node2/c2
dab080b33e76 busybox "top" 18 seconds ago Up 18 seconds swarm-node1/c1
接下来，我们查看net1网络的具体详情：
$ docker network inspect net1
[
{
"Name": "net1",
"Id": "d6a8a2229848d40ce446f8f850a0e713a6c88a9b43583cc463f437f306724f28",
"Created": "2018-05-08T09:21:42.408840755Z",
"Scope": "global",
"Driver": "overlay",
"EnableIPv6": false,
"IPAM": {
"Driver": "default",
"Options": {},
"Config": [
{
"Subnet": "10.0.0.0/24",
"Gateway": "10.0.0.1"
}
]
},
"Internal": false,
"Attachable": false,
"Ingress": false,
"ConfigFrom": {
"Network": ""
},
"ConfigOnly": false,
"Containers": {
"583fde42182a7e8f27527d5c55163a32102dba566ebe1f13d1951ac214849c8d": {
"Name": "c2",
"EndpointID": "b7fcb0039ab21ff06b36ef9ba008c324fabf24badfe45dfa6a30db6763716962",
"MacAddress": "",
"IPv4Address": "10.0.0.3/24",
"IPv6Address": ""
},
"dab080b33e76af0e4c71c9365a6e57b2191b7aacd4f715ca11481403eccce12a": {
"Name": "c1",
"EndpointID": "8a80a83230edfdd9921357f08130fa19ef0b917bc4426aa49cb8083af9edb7f6",
"MacAddress": "",
"IPv4Address": "10.0.0.2/24",
"IPv6Address": ""
}
},
"Options": {},
"Labels": {}
}
]
可以看到，我们刚刚创建的两个容器信息（包含 IP 地址），也在里面。
然后我们测试下两个容器是否可以相互访问（直接 ping 容器名称，也可以访问）：
$ docker exec c1 ping -c 3 10.0.0.3
PING 10.0.0.3 (10.0.0.3): 56 data bytes
64 bytes from 10.0.0.3: seq=0 ttl=64 time=0.903 ms
64 bytes from 10.0.0.3: seq=1 ttl=64 time=0.668 ms
64 bytes from 10.0.0.3: seq=2 ttl=64 time=0.754 ms
--- 10.0.0.3 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 0.668/0.775/0.903 ms
$ docker exec c2 ping -c 3 10.0.0.2
PING 10.0.0.2 (10.0.0.2): 56 data bytes
64 bytes from 10.0.0.2: seq=0 ttl=64 time=0.827 ms
64 bytes from 10.0.0.2: seq=1 ttl=64 time=0.702 ms
64 bytes from 10.0.0.2: seq=2 ttl=64 time=0.676 ms
--- 10.0.0.2 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 0.676/0.735/0.827 ms
$ docker exec c2 ping -c 3 c1
PING c1 (10.0.0.2): 56 data bytes
64 bytes from 10.0.0.2: seq=0 ttl=64 time=1.358 ms
64 bytes from 10.0.0.2: seq=1 ttl=64 time=0.663 ms
64 bytes from 10.0.0.2: seq=2 ttl=64 time=0.761 ms
--- c1 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max = 0.663/0.927/1.358 ms
另附 Docker Swarm 架构图：
参考资料：