基本流程为:
- 做image 并且push到private registry
- 创建replication controller
- 创建service
- 搭建nginx代理
- 测试 下面详述:
1. 制作image
java基础image:
FROM dockerimages.yinnut.com:15043/centos:7
MAINTAINER xuelun-infra morgan.wu@yinnnut.com
ADD jdk-8u60-linux-x64.tar.gz /usr/local/
ENV JAVA_HOME /usr/local/jdk1.8.0_60
ENV PATH $JAVA_HOME/bin:$PATH
Friend-Service镜像:
FROM dockerimages.yinnut.com:15043/yinnut-java:0.1
MAINTAINER xuelun-infra morgan.wu@yinnnut.com
ADD FriendService.war /
ADD jetty-runner.jar /
WORKDIR /
VOLUME ["/var/log"]
2. 创建Replication Controller
rc.yaml文件, 创建了2个friend的pod,设置replicas为2
apiVersion: v1
kind: ReplicationController
metadata:
name: friend-service
spec:
replicas: 2
template:
metadata:
labels:
yinnut-svc: "friend"
version: "0.1"
spec:
containers:
- name: friend-service
image: dockerimages.yinnut.com:15043/friend-service:0.1
ports:
- containerPort: 9999
protocol: TCP
command: ["java"]
args: ["-jar","jetty-runner.jar","--port","9999","--log","jetty.log","FriendService.war"]
##3. 创建Service## svc.yaml文件, 监听本地物理机的 127.0.0.1:30001端口
apiVersion: v1
kind: Service
metadata:
name: friendsvc
labels:
yinnut-svc: friend
spec:
type: NodePort
ports:
- port: 9999
protocol: TCP
targetPort: 9999
name: http
nodePort: 30001
selector:
yinnut-svc: friend
##4. 搭建Nginx反向代理## 安装:
yum install -y http://nginx.org/packages/centos/7/noarch/RPMS/nginx-release-centos-7-0.el7.ngx.noarch.rpm
yum install nginx -y
配置 /etc/nginx/conf.d/default.conf
server {
listen 30000;
server_name localhost;
location / {
port_in_redirect on;
proxy_pass http://127.0.0.1:30001;
proxy_redirect off;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
}
}
设置worker数量为10,并且把30000端口添加到防火墙例外。 访问 http://192.168.1.221:30000端口即可。
##5. 测试## 可以看到有两个Friend的Pod
[root@centos7-node-221 ~]$ kubectl get po
NAME READY STATUS RESTARTS AGE
busybox 1/1 Running 296 12d
friend-service-38riq 1/1 Running 0 6h
friend-service-nn0qt 1/1 Running 0 11m
发送4个请求。其实我发送了上千个请求,怕这里写不下了。
[root@centos7-node-221 ~]$ for i in {1..4}; do time curl http://192.168.1.221:30000/user/1/friend; done
{"rc":0,"val":[{"friendUserId":2,"friendUserName":"who","close":false}]}
real 0m0.029s
user 0m0.004s
sys 0m0.003s
{"rc":0,"val":[{"friendUserId":2,"friendUserName":"who","close":false}]}
real 0m0.039s
user 0m0.001s
sys 0m0.004s
{"rc":0,"val":[{"friendUserId":2,"friendUserName":"who","close":false}]}
real 0m0.028s
user 0m0.002s
sys 0m0.004s
{"rc":0,"val":[{"friendUserId":2,"friendUserName":"who","close":false}]}
real 0m0.035s
user 0m0.003s
sys 0m0.002s
2个Pod分别2个Response,多副本的RR负载均衡工作的很棒。
# 第一个Pod
172.16.91.0 - - [11/Oct/2015:12:33:51 +0000] "GET //192.168.1.221/user/1/friend HTTP/1.0" 200 72
172.16.91.0 - - [11/Oct/2015:12:33:51 +0000] "GET //192.168.1.221/user/1/friend HTTP/1.0" 200 72
# 和第二个Pod
172.16.91.1 - - [11/Oct/2015:12:33:51 +0000] "GET //192.168.1.221/user/1/friend HTTP/1.0" 200 72
172.16.91.1 - - [11/Oct/2015:12:33:51 +0000] "GET //192.168.1.221/user/1/friend HTTP/1.0" 200 72
6. 总结##
Kubernetes棒棒哒!!! ;-) 负载均衡棒棒哒!