与docker搭建hadoop不一样的是，本次采用Dockerfile的方式构建镜像，然后启动镜像，形成多主机多容器的方式启动Hadoop集群。
1.在各个宿主机上搭建网桥weave
curl -L git.io/weave -o /usr/local/bin/weave
weave launch
2.各个宿主机通过weave网桥链接
192.168.130.166主机上操作：
weave connect 192.168.130.167
weave connect 192.168.130.168
192.168.130.167主机上操作：
weave connect 192.168.130.166
weave connect 192.168.130.168
192.168.130.168主机上操作：
weave connect 192.168.130.166
weave connect 192.168.130.167
3.各个宿主机上创建子网段，使得同一网段的hadoop容器节点能互相通信：
docker network create hadoop
4.创建目录hadoop-cluster-docker，并在该目录下，写入hadoop的配置相关：
相关数据上传到git中，详情：https://github.com/mayunzhen/hadoop2.8.5-cluster-docker
5.拉取git项目:
git clone https://github.com/mayunzhen/hadoop2.8.5-cluster-docker.git
cd hadoop2.8.5-cluster-docker
./build-image.sh
6.启动集群：
随意哪个宿主机上随意启动(tips:weave固定的IP必须在同一个网段中，以便同一网段的容器能通信)：
docker run -itd -h hadoop-master --name hadoop-master --net=hadoop -v /etc/localtime:/etc/localtime:ro -p 50070:50070 -p 8088:8088 -p 9000:9000 iammayunzhen/hadoop:2.8.5
weave attach 192.168.1.10/24 hadoop-master
docker run -itd -h hadoop-slave1 --name hadoop-slave1 --net=hadoop -v /etc/localtime:/etc/localtime:ro centos_hdp:2.8.5
weave attach 192.168.1.11/24 hadoop-slave1
docker run -itd -h hadoop-slave2 --name hadoop-slave2 --net=hadoop -v /etc/localtime:/etc/localtime:ro centos_hdp:2.8.5
weave attach 192.168.1.12/24 hadoop-slave2
7.查看集群状态
。。。
原文地址：https://www.cnblogs.com/yunger/p/12067883.html