实现多机容器的通信，首先需要保证容器的ip在多台虚拟机中都是唯一的。以上图为例，在192.168.205.10中新建容器，就不能再出现容器的ip为172.17.0.3了，因为这个ip已经在192.168.205.10中存在了。
etcd安装
为了保证容器在多机中的ip唯一，可以使用etcd，etcd是一个开源的、分布式的键值对数据存储系统
https://coreos.com/etcd/
在docker-node1上安装etcd
wget https://github.com/coreos/etcd/releases/download/v3.0.12/etcd-v3.0.12-linux-amd64.tar.gz
tar zxvf etcd-v3.0.12-linux-amd64.tar.gz
cd etcd-v3.0.12-linux-amd64
nohup ./etcd --name docker-node1 --initial-advertise-peer-urls http://192.168.205.10:2380
--listen-peer-urls http://192.168.205.10:2380
--listen-client-urls http://192.168.205.10:2379,http://127.0.0.1:2379
--advertise-client-urls http://192.168.205.10:2379
--initial-cluster-token etcd-cluster
--initial-cluster docker-node1=http://192.168.205.10:2380,docker-node2=http://192.168.205.11:2380
--initial-cluster-state new &在docker-node2上安装etcd
wget https://github.com/coreos/etcd/releases/download/v3.0.12/etcd-v3.0.12-linux-amd64.tar.gz
tar zxvf etcd-v3.0.12-linux-amd64.tar.gz
cd etcd-v3.0.12-linux-amd64/
nohup ./etcd --name docker-node2 --initial-advertise-peer-urls http://192.168.205.11:2380
--listen-peer-urls http://192.168.205.11:2380
--listen-client-urls http://192.168.205.11:2379,http://127.0.0.1:2379
--advertise-client-urls http://192.168.205.11:2379
--initial-cluster-token etcd-cluster
--initial-cluster docker-node1=http://192.168.205.10:2380,docker-node2=http://192.168.205.11:2380
--initial-cluster-state new &在docker-node1或docker-node2任一主机上检查cluster运行状态
./etcdctl cluster-health
member 21eca106efe4caee is healthy: got healthy result from http://192.168.205.10:2379
member 8614974c83d1cc6d is healthy: got healthy result from http://192.168.205.11:2379
cluster is healthydocker重启
使用 systemctl stop docker 命令，关闭docker服务后再使用如下方式重启
docker-node1
sudo /usr/bin/dockerd -H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock --cluster-store=etcd://192.168.205.10:2379 --cluster-advertise=192.168.205.10:2375 &docker-node2
sudo /usr/bin/dockerd -H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock --cluster-store=etcd://192.168.205.11:2379 --cluster-advertise=192.168.205.11:2375 &创建overlay network
在docker-node1上创建一个demo的overlay network
docker network create -d overlay demodocker-node1，可以看到demo的DRIVER是overlay，SCOPE是global
[vagrant@docker-node1 etcd-v3.0.12-linux-amd64]$ docker network ls
NETWORK ID NAME DRIVER SCOPE
9bacf5670fa1 bridge bridge local
5a8cd36174a1 demo overlay global
efb6975c8935 host host local
4213425c5293 my-bridge bridge local
3fa0f2e1a00b none null local然后在docker-node2上发现会同步创建
[vagrant@docker-node2 etcd-v3.0.12-linux-amd64]$ docker network ls
NETWORK ID NAME DRIVER SCOPE
1c79331058ff bridge bridge local
5a8cd36174a1 demo overlay global
148fe990ddc5 host host local
452bd665cf13 none null local通过查看etcd的key-value，发现docker network inspect demo的内容与key-value中的内容是一样的，也就是所谓的分布式网络
[
{
"Name": "demo",
"Id": "5a8cd36174a1b37f9146aaa99de33ff20bfa8df472e07412e36048a39fe9a0e9",
"Created": "2018-07-03T09:21:08.95776633Z",
"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": {},
"Options": {},
"Labels": {}
}
]测试多机通信效果
在docker-node1上
docker run -d --name test1 --network demo busybox /bin/sh -c "while true;do sleep 3600;done"在docker-node2上如果重复执行上述命令，就会报如下错误了，从另一方面也说明了etcd在发挥作用
docker: Error response from daemon: Conflict. The container name "/test1" is already in use by container "14d690ada81a4510eb5dd24286d061064e8131f549dcb5320f5b5b441cc49c40". You have to remove (or rename) that container to be able to reuse that name.可以改为
docker run -d --name test2 --network demo busybox /bin/sh -c "while true;do sleep 3600;done"此时查看一下docker-node1与docker-node2分别对应的test1与test2容器的网络
docker network inspect demo
{
"Containers": {
"a9fa503852b26267d13b67067ba6c119eebcf0677f8365483b3b798dcab616a5": {
"Name": "test1",
"EndpointID": "3869243fcfb6a975fadf64dc31533161d2922137b18af0c3ee78c02f82b84def",
"MacAddress": "",
"IPv4Address": "10.0.0.2/24",
"IPv6Address": ""
},
"ep-8f8e30e36867c5b906bc84438049b2911660057fe031536a97efcf01d7d98699": {
"Name": "test2",
"EndpointID": "8f8e30e36867c5b906bc84438049b2911660057fe031536a97efcf01d7d98699",
"MacAddress": "",
"IPv4Address": "10.0.0.3/24",
"IPv6Address": ""
}
}
}然后在test1上ping位于docker-node2上的test2容器是可以ping通的
docker exec test1 ping 10.0.0.3在test2上ping位于docker-node1上的test1容器也是可以ping通的
docker exec test2 ping 10.0.0.2成功实现了多机容器通信的效果