yum方式安装部署k8s集群

安装前准备：

环境：

主机	服务	ip
k8s-master	etcd、	10.0.0.11
k8s-node-1	k8s-node	10.0.0.12
k8s-node-2	k8s-node	10.0.0.13

所有主机配置hosts解析

10.0.0.11     k8s-master
10.0.0.12     k8s-node-1
10.0.0.13     k8s-node-2

关闭防火墙和selinux

systemctl stop firewalld
setenforce 0

yum源使用的是阿里云的

curl -o /etc/yum.repos.d/CentOS-Base.repo http://mirrors.aliyun.com/repo/Centos-7.repo

kubernetes架构图：

一、首先配置k8s-master

1.k8s-master安装etcd：

yum install -y etcd

2.配置etcd

vim /etc/etcd/etcd.conf
#修改下面三项
ETCD_LISTEN_CLIENT_URLS="http://0.0.0.0:2379"
ETCD_NAME="k8s-master"
ETCD_ADVERTISE_CLIENT_URLS="http://10.0.0.11:2379"


#启动服务
systemctl enable etcd
systemctl start etcd


#测试，正常设置和取值
[root@k8s-master ~]# etcdctl set test/k v
v
[root@k8s-master ~]# etcdctl get test/k
v


#查看etcd健康状态
[root@k8s-master ~]# etcdctl -C http://10.0.0.11:2379 cluster-health
member 8e9e05c52164694d is healthy: got healthy result from http://10.0.0.11:2379
cluster is healthy

 

3.安装kubernetes-master

yum install -y kubernetes-master

 

4.配置apiserver

vim /etc/kubernetes/apiserver
# 修改下面五项
KUBE_API_ADDRESS="--insecure-bind-address=0.0.0.0"
KUBE_API_PORT="--port=8080"
KUBELET_PORT="--kubelet-port=10250"
KUBE_ETCD_SERVERS="--etcd-servers=http://10.0.0.11:2379"
KUBE_ADMISSION_CONTROL="--admission-control=NamespaceLifecycle,NamespaceExists,LimitRanger,SecurityContextDeny,ResourceQuota"

 

5.指定apiserver地址，给controller-manager和scheduler使用

vim /etc/kubernetes/config
KUBE_MASTER="--master=http://10.0.0.11:8080"

 

6.启动服务

systemctl enable kube-apiserver
systemctl enable kube-controller-manager
systemctl enable kube-scheduler

systemctl restart kube-apiserver
systemctl restart kube-controller-manager
systemctl restart kube-scheduler

 

7.检查其他组件的状态

[root@k8s-master ~]# kubectl get componentstatus 
NAME                 STATUS       MESSAGE            ERROR
etcd-0               Healthy      {"health":"true"} 
scheduler            Healthy ok 
controller-manager   Healthy ok

 

二、配置node节点(node2也是同样配置)

1.安装kubernetes-node

yum install -y kubernetes-node

 

2.修改kubelet，注意hostname，每个节点需要不同的名字

vim /etc/kubernetes/kubelet

KUBELET_ADDRESS="--address=0.0.0.0"
KUBELET_PORT="--port=10250"
KUBELET_HOSTNAME="--hostname-override=k8s-node-1"
KUBELET_API_SERVER="--api-servers=http://10.0.0.11:8080"

 

3.指定apiserver地址，给kube-proxy使用

vim /etc/kubernetes/config
KUBE_MASTER="--master=http://10.0.0.11:8080"

4.启动服务

systemctl enable kubelet
systemctl enable kube-proxy

systemctl restart kubelet.service 
systemctl restart kube-proxy.service

 

5. 在master上查看node节点

kubectl get nodes
NAME       STATUS    AGE
k8s-node-1 Ready     1m
k8s-node-2 Ready     3s

 

至此k8s已经装完了，接下来配置不同宿主机之间容器的通信，这里使用的flannel。

安装配置flannel

1.所有节点安装flannel

yum install -y flannel
systemctl enable flanneld

 

2.修改fannel配置,三台机器都做此操作

vim /etc/sysconfig/flanneld
FLANNEL_ETCD_ENDPOINTS="http://10.0.0.11:2379"

#也可以使用sed替换
sed -i s#127.0.0.1:2379#10.0.0.11:2379#g /etc/sysconfig/flanneld

 

3.通过etcd配置ip地址范围（key:value形式）

etcdctl mk /atomic.io/network/config '{"Network":"172.16.0.0/16"}'

 

4.重启服务

master节点需要自行安装docker，配置完flannel也需要重启docker。

master节点：

systemctl restart flanneld
systemctl restart kube-apiserver
systemctl restart kube-controller-manager 
systemctl restart kube-scheduler

node节点：

systemctl restart flanneld
systemctl restart docker
systemctl restart kubelet
systemctl restart kube-proxy

 

docker1.13版本中的一个问题，需要修改iptables，否则容器之间不通。

vim /usr/lib/systemd/system/docker.service
ExecStartPost=/sbin/iptables -P FORWARD ACCEPT
systemctl daemon-reload
systemctl restart docker