更合理的部署方式参见《Centos下Kubernetes+Flannel部署（新）》
一、准备工作
1. 三台centos主机
k8s（即kubernetes，下同）master: 10.16.42.200
k8s node1: 10.16.42.198
k8s node2: 10.16.42.199
2. 程序下载（百度网盘）：k8s-1.0.1(也可以使用测试版k8s-1.1.2.beta)，Docker-1.8.2，cadvisor-0.14.0，etcd-2.2.1，flannel-0.5.5
二、ETCD集群部署
分别向三台主机的/etc/hosts文件中追加如下设置：
10.16.42.198 bx--
10.16.42.199 bx--
10.16.42.200 bx--
在三台主机中分别解压etcd.tar，将其中的 etcd 和 etcdctl 复制到你的工作目录(如 /openxxs/bin，下同)。
在200的/openxxs/bin目录下创建脚本start_etcd.sh并执行：
#!/bin/bash etcd_token=kb2-etcd-cluster
local_name=kbetcd0
local_ip=10.16.42.200
local_peer_port=
local_client_port1=
local_client_port2=
node1_name=kbetcd1
node1_ip=10.16.42.198
node1_port=
node2_name=kbetcd2
node2_ip=10.16.42.199
node2_port= ./etcd -name $local_name \
-initial-advertise-peer-urls http://$local_ip:$local_peer_port \
-listen-peer-urls http://0.0.0.0:$local_peer_port \
-listen-client-urls http://0.0.0.0:$local_client_port1,http://0.0.0.0:$local_client_port2 \
-advertise-client-urls http://$local_ip:$local_client_port1,http://$local_ip:$local_client_port2 \
-initial-cluster-token $etcd_token \
-initial-cluster $local_name=http://$local_ip:$local_peer_port,$node1_name=http://$node1_ip:$node1_port,$node2_name=http://$node2_ip:$node2_port \
-initial-cluster-state new
在198的/openxxs/bin目录下创建脚本start_etcd.sh并执行：
#!/bin/bash etcd_token=kb2-etcd-cluster
local_name=kbetcd1
local_ip=10.16.42.198
local_peer_port=
local_client_port1=
local_client_port2=
node1_name=kbetcd0
node1_ip=10.16.42.200
node1_port=
node2_name=kbetcd2
node2_ip=10.16.42.199
node2_port= ./etcd -name $local_name \
-initial-advertise-peer-urls http://$local_ip:$local_peer_port \
-listen-peer-urls http://0.0.0.0:$local_peer_port \
-listen-client-urls http://0.0.0.0:$local_client_port1,http://0.0.0.0:$local_client_port2 \
-advertise-client-urls http://$local_ip:$local_client_port1,http://$local_ip:$local_client_port2 \
-initial-cluster-token $etcd_token \
-initial-cluster $local_name=http://$local_ip:$local_peer_port,$node1_name=http://$node1_ip:$node1_port,$node2_name=http://$node2_ip:$node2_port \
-initial-cluster-state new &
在199的/openxxs/bin目录下创建脚本start_etcd.sh并执行：
#!/bin/bash etcd_token=kb2-etcd-cluster
local_name=kbetcd2
local_ip=10.16.42.199
local_peer_port=
local_client_port1=
local_client_port2=
node1_name=kbetcd1
node1_ip=10.16.42.198
node1_port=
node2_name=kbetcd0
node2_ip=10.16.42.200
node2_port= ./etcd -name $local_name \
-initial-advertise-peer-urls http://$local_ip:$local_peer_port \
-listen-peer-urls http://0.0.0.0:$local_peer_port \
-listen-client-urls http://0.0.0.0:$local_client_port1,http://0.0.0.0:$local_client_port2 \
-advertise-client-urls http://$local_ip:$local_client_port1,http://$local_ip:$local_client_port2 \
-initial-cluster-token $etcd_token \
-initial-cluster $local_name=http://$local_ip:$local_peer_port,$node1_name=http://$node1_ip:$node1_port,$node2_name=http://$node2_ip:$node2_port \
-initial-cluster-state new &
在各主机上执行类似如下命令查看etcd是否正常运行：
curl -L http://10.16.42.198:4012/version
curl -L http://10.16.42.199:4012/version
curl -L http://10.16.42.200:4012/version
如果返回值均为 “{"etcdserver":"2.2.1","etcdcluster":"2.2.0"}” 说明ETCD部署成功。
三、Docker安装与设置
yum install docker-engine-1.8.-.el7.centos.x86_64.rpm -y
各个主机上安装成功后，修改 /etc/sysconfig/docker 文件为：
OPTIONS="-g /opt/scs/docker --insecure-registry 10.11.150.76:5000"
其中的--insecure-registry表示使用自己私有的镜像仓库。
修改 /lib/systemd/system/docker.service 内容为：
[Unit]
Description=Docker Application Container Engine
Documentation=https://docs.docker.com
After=network.target docker.socket
Requires=docker.socket [Service]
Type=notify
EnvironmentFile=/etc/sysconfig/docker
ExecStart=/usr/bin/docker -d $OPTIONS \
$DOCKER_STORAGE_OPTIONS \
$DOCKER_NETWORK_OPTIONS \
$ADD_REGISTRY \
$BLOCK_REGISTRY \
$INSECURE_REGISTRY
#ExecStart=/usr/bin/docker daemon -H fd://
MountFlags=slave
LimitNOFILE=
LimitNPROC=
LimitCORE=infinity [Install]
WantedBy=multi-user.target
注意，k8s会托管你的docker，如果之前在主机上用docker创建或运行了一些容器，注意数据的备份。
四、Flannel安装与设置
yum localinstall flannel-0.5.-.fc24.x86_64.rpm
各个主机上安装成功后，修改 /etc/sysconfig/flanneld 内容为：
# Flanneld configuration options # etcd url location. Point this to the server where etcd runs
FLANNEL_ETCD="http://10.16.42.200:4012" # etcd config key. This is the configuration key that flannel queries
# For address range assignment
FLANNEL_ETCD_KEY="/coreos.com/network" # Any additional options that you want to pass
#FLANNEL_OPTIONS=""
即确定flanneld使用的etcd服务地址和etcd中存储flannel相关设置的key值。
修改 /lib/systemd/system/flanneld.service 内容为：
[Unit]
Description=Flanneld overlay address etcd agent
After=network.target
After=network-online.target
Wants=network-online.target
After=etcd.service
Before=docker.service [Service]
Type=notify
EnvironmentFile=/etc/sysconfig/flanneld
EnvironmentFile=-/etc/sysconfig/docker-network
ExecStart=/usr/bin/flanneld -etcd-endpoints=${FLANNEL_ETCD} -etcd-prefix=${FLANNEL_ETCD_KEY} $FLANNEL_OPTIONS
ExecStartPost=/usr/libexec/flannel/mk-docker-opts.sh -k DOCKER_NETWORK_OPTIONS -d /run/flannel/docker
Restart=on-failure [Install]
WantedBy=multi-user.target
RequiredBy=docker.service
五、具体部署过程
1. 启动ETCD
ETCD是k8s正常运行的基础，因此按照第二步中的方式运行脚本并测试部署成功后再进行其它程序的启动。
2. 启动Flannel
启动Flannel前停止docker、iptables和firewall服务的运行：
systemctl stop docker
systemctl disable iptables-services firewalld
systemctl stop iptables-services firewalld
使用 ps aux | grep docker 查看 docker 是不是以 daemon 的形式运行着。如果是，kill 掉该进程。
使用 ifconfig 查看是否存在 docker0 及 flannel 相关的网桥。如果有，使用 ip link delete docker0 删除。
以上准备工作做好后，还需要往ETCD中写入flannel的相关配置，即在 200 主机上创建 flannel-config.json 文件，内容为：
{
"Network": "172.16.0.0/12",
"SubnetLen": ,
"Backend": {
"Type": "vxlan",
"VNI":
}
}
即规定了flannel可用的子网段和网络包封装方式，然后将其写入ETCD中（注意这里的key值和Flannel启动的FLANNEL_ETCD_KEY参数值保持一致）：
./etcdctl --peers=http://10.16.42.200:4012 set /coreos.com/network/config < flannel-config.json
然后在各个主机上启动flannel：
systemctl start flanneld
3. 启动docker
在各个主机上启动docker服务：
systemctl start docker
然后使用ifconfig查看docker0和flannel.1的IP网段，如果flannel.1的网段包含了docker0的网段，则说明flannel的配置和启动是没问题的。
4. 启动master上的k8s服务
./kube-apiserver --logtostderr=true --v= --etcd_servers=http://127.0.0.1:4012 --kubelet_port=10250 --allow_privileged=false --service-cluster-ip-range=172.16.0.0/12 --insecure-bind-address=0.0.0.0 --insecure-port=8080 & ./kube-controller-manager --logtostderr=true --v= --master=http://bx-42-200:8080 --cloud-provider="" & ./kube-scheduler --logtostderr=true --v= --master=http://bx-42-200:8080 &
注意在启动 kube-controller-manager 时可能会报如下错误：
plugins.go:] No cloud provider specified.
controllermanager.go:] Failed to start service controller: ServiceController should not be run without a cloudprovider.
这是由 --cloud-provider 的值为空或未指定该参数造成的，但对整体的k8s运行无太大影响，所以可以忽略（该bug参见github讨论：戳这里）。
5. 启动node上的k8s服务
./kube-proxy --logtostderr=true --v= --master=http://bx-42-200:8080 --proxy-mode=iptables & ./kubelet --logtostderr=true --v= --api_servers=http://bx-42-200:8080 --address=0.0.0.0 --allow_privileged=false --pod-infra-container-image=10.11.150.76:5000/kubernetes/pause:latest &
注意这里的 --pod-infra-container-image 参数设置。每个pod启动时都要先启动一个/kubernetes/pause:latest容器来进行一些基本的初始化工作，而默认该镜像的下载地址为 gcr.io/google_containers/pause:0.8.0 。由于GWF的存在可能会连接不上该资源，所以可以将该镜像下载下来之后再push到自己的docker本地仓库中，启动 kubelet 时从本地仓库中读取即可。
还有注意 --proxy-mode=iptables 参数是在k8s 1.1实验版本中才有的，其含义的官方解释如下：
--proxy-mode="": Which proxy mode to use: 'userspace' (older, stable) or 'iptables' (experimental). If blank, look at the Node object on the Kubernetes API and respect the 'net.experimental.kubernetes.io/proxy-mode' annotation if provided. Otherwise use the best-available proxy (currently userspace, but may change in future versions). If the iptables proxy is selected, regardless of how, but the system's kernel or iptables versions are insufficient, this always falls back to the userspace proxy.
如果不支持 --proxy-mode=iptables 则会报类似如下错误：
W1119 ::12.187930 server.go:] Failed to start in resource-only container "/kube-proxy": write /sys/fs/cgroup/memory/kube-proxy/memory.swappiness: invalid argument
E1119 ::12.198572 proxier.go:] Error removing userspace rule: error checking rule: exit status : iptables v1.4.21: Couldn't load target `KUBE-PORTALS-HOST':No such file or directory Try `iptables -h' or 'iptables --help' for more information.
E1119 ::12.200286 proxier.go:] Error removing userspace rule: error checking rule: exit status : iptables v1.4.21: Couldn't load target `KUBE-PORTALS-CONTAINER':No such file or directory Try `iptables -h' or 'iptables --help' for more information.
E1119 ::12.202162 proxier.go:] Error removing userspace rule: error checking rule: exit status : iptables v1.4.21: Couldn't load target `KUBE-NODEPORT-HOST':No such file or directory Try `iptables -h' or 'iptables --help' for more information.
E1119 ::12.204058 proxier.go:] Error removing userspace rule: error checking rule: exit status : iptables v1.4.21: Couldn't load target `KUBE-NODEPORT-CONTAINER':No such file or directory Try `iptables -h' or 'iptables --help' for more information.
E1119 ::12.205848 proxier.go:] Error flushing userspace chain: error flushing chain "KUBE-PORTALS-CONTAINER": exit status : iptables: No chain/target/match by that name.
E1119 ::12.207467 proxier.go:] Error flushing userspace chain: error flushing chain "KUBE-PORTALS-HOST": exit status : iptables: No chain/target/match by that name.
E1119 ::12.209000 proxier.go:] Error flushing userspace chain: error flushing chain "KUBE-NODEPORT-HOST": exit status : iptables: No chain/target/match by that name.
E1119 ::12.210580 proxier.go:] Error flushing userspace chain: error flushing chain "KUBE-NODEPORT-CONTAINER": exit status : iptables: No chain/target/match by that name.
六、测试
以上部署完成之后，在任意主机上执行以下命令查看结点状态：
./kubectl -s 10.16.42.200: get nodes
如果返回类似如下内容则说明apiserver是正常服务的：
NAME LABELS STATUS AGE
bx-- kubernetes.io/hostname=bx-- Ready 1d
bx-- kubernetes.io/hostname=bx-- Ready 1d
创建 test.yaml 文件，内容如下：
apiVersion: v1
kind: ReplicationController
metadata:
name: test-
spec:
replicas:
template:
metadata:
labels:
app: test-
spec:
containers:
- name: iperf
image: 10.11.150.76:/openxxs/iperf:1.2
nodeSelector:
kubernetes.io/hostname: bx--
---
apiVersion: v1
kind: ReplicationController
metadata:
name: test-
spec:
replicas:
template:
metadata:
labels:
app: test-
spec:
containers:
- name: iperf
image: 10.11.150.76:/openxxs/iperf:1.2
nodeSelector:
kubernetes.io/hostname: bx--
---
apiVersion: v1
kind: ReplicationController
metadata:
name: test-
spec:
replicas:
template:
metadata:
labels:
app: test-
spec:
containers:
- name: iperf
image: 10.11.150.76:/openxxs/iperf:1.2
nodeSelector:
kubernetes.io/hostname: bx--
---
apiVersion: v1
kind: ReplicationController
metadata:
name: test-
spec:
replicas:
template:
metadata:
labels:
app: test-
spec:
containers:
- name: iperf
image: 10.11.150.76:/openxxs/iperf:1.2
nodeSelector:
kubernetes.io/hostname: bx--
表示在198上创建 test-1 和 test-2 两个pod，在199上创建 test-3 和 test-4 两个pod。注意其中的 image 等参数根据实际情况进行修改。
通过test.yaml创建pods：
./kubectl -s 10.16.42.200: create -f test.yaml
通过 get pods 查看pods的创建和运行状态：
./kubectl -s 10.16.42.200: get pods
如果创建成功并正常运行则会显示类似如下内容：
NAME READY STATUS RESTARTS AGE
test--a9dn3 / Running 1d
test--64urt / Running 1d
test--edt2l / Running 1d
test--l6egg / Running 1d
在198上通过 docker exec 进入test-2对应的容器，通过 ip addr show 查看IP；同样在199上进入test-4对应的容器查看IP。然后在 test-2和 test-4 容器中互相ping 对方的IP，如果ping通了，说明flannel也正常工作了。