在Ubuntu16.04集群上手工部署Kubernetes

目前Kubernetes为Ubuntu提供的kube-up脚本，不支持15.10以及16.04这两个使用systemd作为init系统的版本。

这里详细介绍一下如何以非Docker方式在Ubuntu16.04集群上手动安装部署Kubernetes的过程。

环境信息

 

组件	版本
etcd	2.3.1
Flannel	0.5.5
kubernetes	1.5.1

主机信息

 

主机	IP	OS
k8s-master	172.12.24.36	ubuntu 16.04
k8s-node01	172.12.24.37	ubuntu 16.04
k8s-node02	172.12.24.38	ubuntu 16.04

 

安装docker

sudo apt-get install docker.io

安装Go

sudo apt-get install golang-go

部署etcd集群

我们将在3台机器上安装etcd集群

1.在部署机器上下载etcd

ETCD_VERSION=${ETCD_VERSION:-"2.3.1"}

ETCD="etcd-v${ETCD_VERSION}-Linux-amd64"

curl -L   https://github.com/coreos/etcd/releases/download/v${ETCD_VERSION}/${ETCD}.tar.gz -o etcd.tar.gz

2.解压etcd到tmp目录

tar xzf etcd.tar.gz -C /tmp

3.进入解压目录

cd /tmp/etcd-v${ETCD_VERSION}-linux-amd64

4.创建etcd集群

for h in k8s-master k8s-node01 k8s-node02; do ssh user@$h mkdir -p '$HOME/kube' && scp -r etcd* user@$h:~/kube; done
for h in k8s-master k8s-node01 k8s-node02; do ssh user@$h 'sudo mkdir -p /opt/bin && sudo mv $HOME/kube/* /opt/bin && rm -rf $home/kube/*'; done

注意：

1、这种创建集群的方式，每台集群的登录用户必须一致，如user@$h中的user指当前机器的登录用户。

2、sudo方式必须是免密码方式，否则执行当中报错。

修改为免密码方式：

sudo chmod   u+w  /etc/sudoers

sudo vim   /etc/sudoers

修改sudoers文件中的内容

%sudo    ALL=(ALL:ALL) ALL

修改为

%sudo    ALL=(ALL:ALL) NOPASSWD:ALL

3、该设置只需要在master节点执行，其他节点将会配置好

配置etcd服务

在每台主机上，分别创建/opt/config/etcd.conf和/lib/systemd/system/etcd.service文件，（注意修改红色粗体处的IP地址）

1.创建/opt/config/etcd.conf文件

sudo mkdir -p /var/lib/etcd/

sudo mkdir -p /opt/config/

sudo  cat <<EOF |  sudo tee /opt/config/etcd.conf
ETCD_DATA_DIR=/var/lib/etcd
ETCD_NAME=$(hostname)
ETCD_INITIAL_CLUSTER=k8s-master=http://172.12.24.36:2380,k8s-node01=http://172.12.24.37:2380,k8s-node02=http://172.12.24.38:2380
ETCD_INITIAL_CLUSTER_STATE=new
ETCD_LISTEN_PEER_URLS=http://172.12.24.36:2380
ETCD_INITIAL_ADVERTISE_PEER_URLS=http://172.12.24.36:2380
ETCD_ADVERTISE_CLIENT_URLS=http://172.12.24.36:2379
ETCD_LISTEN_CLIENT_URLS=http://172.12.24.36:2379
GOMAXPROCS=$(nproc)
EOF

2.创建 /lib/systemd/system/etcd.service文件

sudo  cat <<EOF |  sudo tee /lib/systemd/system/etcd.service
[Unit]
Description=Etcd Server
Documentation=https://github.com/coreos/etcd
After=network.target


[Service]
User=root
Type=simple
EnvironmentFile=-/opt/config/etcd.conf
ExecStart=/opt/bin/etcd
Restart=on-failure
RestartSec=10s
LimitNOFILE=40000

[Install]
WantedBy=multi-user.target
EOF

每台机器重新加载启动etcd服务

sudo systemctl daemon-reload

sudo systemctl enable etcd

sudo systemctl start etcd

下载Flannel

1.设定版本变量

FLANNEL_VERSION=${FLANNEL_VERSION:-"0.5.5"}

2.下载并重命名

curl -L  https://github.com/coreos/flannel/releases/download/v${FLANNEL_VERSION}/flannel-${FLANNEL_VERSION}-linux-amd64.tar.gz     flannel.tar.gz

3.解压到tmp目录

tar xzf  flannel.tar.gz -C /tmp

编译K8s

1.通过Git克隆kubernetes源码，注意需要克隆的分支

git clone -b  v1.5.1  https://github.com/kubernetes/kubernetes.git

2.进入源码目录

cd kubernetes

3.编译打包源码

make release-skip-tests

4.将编译打包完成的包解压到tmp目录

tar xzf _output/release-stage/full/kubernetes/server/kubernetes-server-linux-amd64.tar.gz  -C   /tmp

Note
除了linux/amd64，默认还会为其他平台做交叉编译。为了减少编译时间，可以修改hack/lib/golang.sh，把KUBE_SERVER_PLATFORMS， KUBE_CLIENT_PLATFORMS和KUBE_TEST_PLATFORMS中除linux/amd64以外的其他平台注释掉。

部署K8s Master

1.切换到tmp目录

cd /tmp

2.将tmp中的文件远程复制到kubernetes相关文件目录中

scp kubernetes/server/bin/kube-apiserver  kubernetes/server/bin/kube-controller-manager  kubernetes/server/bin/kube-scheduler kubernetes/server/bin/kubelet kubernetes/server/bin/kube-proxy user@172.12.24.36:~/kube

3.将tmp中的文件远程复制到flannel相关的文件目录中

scp flannel-${FLANNEL_VERSION}/flanneld user@172.12.24.36:~/kube

4.通过ssh方式连接服务器，移动文件目录
ssh -t user@172.12.24.36   'sudo mv ~/kube/* /opt/bin/'

注意：user为当前机器的登录用户

创建证书
在master主机上 ，运行如下命令创建证书

1.创建目录

sudo mkdir -p /srv/kubernetes/
2.切换目录
cd /srv/kubernetes
3.设置环境变量
export MASTER_IP=172.12.24.36

4.创建证书

sudo openssl genrsa -out ca.key 2048
sudo openssl req -x509 -new -nodes -key ca.key -subj "/CN=${MASTER_IP}" -days 10000 -out ca.crt
sudo openssl genrsa -out server.key 2048
sudo openssl req -new -key server.key -subj "/CN=${MASTER_IP}" -out server.csr
sudo openssl x509 -req -in server.csr -CA ca.crt -CAkey ca.key -CAcreateserial -out server.crt -days 10000

配置kube-apiserver服务
我们使用如下的Service以及Flannel的网段：
SERVICE_CLUSTER_IP_RANGE=172.18.0.0/16
FLANNEL_NET=192.168.0.0/16
在master主机上，创建/lib/systemd/system/kube-apiserver.service文件，

sudo vim   /lib/systemd/system/kube-apiserver.service

内容如下
[Unit]
Description=Kubernetes API Server
Documentation=https://github.com/kubernetes/kubernetes
After=network.target

[Service]
User=root
ExecStart=/opt/bin/kube-apiserver   \
 --insecure-bind-address=0.0.0.0  \
 --insecure-port=8080  \
 --etcd-servers=http://172.12.24.36:2379, http://172.12.24.37:2379, http://172.12.24.38:2379 \
 --logtostderr=true  \
 --allow-privileged=false  \
 --service-cluster-ip-range=172.18.0.0/16  \
 --admission-control=NamespaceLifecycle,LimitRanger,ServiceAccount,SecurityContextDeny,ResourceQuota  \
 --service-node-port-range=30000-32767   \
 --advertise-address=172.12.24.36   \
 --client-ca-file=/srv/kubernetes/ca.crt   \
 --tls-cert-file=/srv/kubernetes/server.crt   \
 --tls-private-key-file=/srv/kubernetes/server.key 
 
Restart=on-failure
Type=notify
LimitNOFILE=65536

[Install]
WantedBy=multi-user.target

配置kube-controller-manager服务
在master主机上，创建/lib/systemd/system/kube-controller-manager.service文件，

sudo vim /lib/systemd/system/kube-controller-manager.service

内容如下
[Unit]
Description=Kubernetes Controller Manager
Documentation=https://github.com/kubernetes/kubernetes

[Service]
User=root
ExecStart=/opt/bin/kube-controller-manager   \
  --master=127.0.0.1:8080  \
  --root-ca-file=/srv/kubernetes/ca.crt   \
  --service-account-private-key-file=/srv/kubernetes/server.key   \
  --logtostderr=true
Restart=on-failure
LimitNOFILE=65536

[Install]
WantedBy=multi-user.target

配置kuber-scheduler服务
在master主机上，创建/lib/systemd/system/kube-scheduler.service文件，

sudo vim  /lib/systemd/system/kube-scheduler.service

内容如下
[Unit]
Description=Kubernetes Scheduler
Documentation=https://github.com/kubernetes/kubernetes

[Service]
User=root
ExecStart=/opt/bin/kube-scheduler  \
  --logtostderr=true  \
  --master=127.0.0.1:8080
Restart=on-failure
LimitNOFILE=65536

[Install]
WantedBy=multi-user.target

配置flanneld服务
在master主机上，创建/lib/systemd/system/flanneld.service文件，

sudo vim  /lib/systemd/system/flanneld.service

内容如下
[Unit]
Description=Flanneld
Documentation=https://github.com/coreos/flannel
After=network.target
Before=docker.service

[Service]
User=root
ExecStart=/opt/bin/flanneld  \
  --etcd-endpoints="http://172.12.24.36:2379,http://172.12.24.37:2379,http://172.12.24.38:2379"  \
  --iface=172.12.24.36  \
  --ip-masq
Restart=on-failure
Type=notify
LimitNOFILE=65536

启动服务
/opt/bin/etcdctl --endpoints="http://172.12.24.54:2379,http://172.12.24.56:2379,http://172.12.24.60:2379" mk /coreos.com/network/config '{"Network":"192.168.0.0/16", "Backend": {"Type": "vxlan"}}'

sudo systemctl daemon-reload
sudo systemctl enable kube-apiserver
sudo systemctl enable kube-controller-manager
sudo systemctl enable kube-scheduler
sudo systemctl enable flanneld

sudo systemctl start kube-apiserver
sudo systemctl start kube-controller-manager
sudo systemctl start kube-scheduler
sudo systemctl start flanneld

修改Docker服务
source /run/flannel/subnet.env

sudo sed -i "s|^ExecStart=/usr/bin/dockerd -H fd://$|ExecStart=/usr/bin/dockerd -H tcp://127.0.0.1:4243 -H unix:///var/run/docker.sock --bip=${FLANNEL_SUBNET} --mtu=${FLANNEL_MTU}|g" /lib/systemd/system/docker.service
rc=0
ip link show docker0 >/dev/null 2>&1 || rc="$?"
if [[ "$rc" -eq "0" ]]; then
ip link set dev docker0 down
ip link delete docker0
fi


sudo systemctl daemon-reload
sudo systemctl enable docker
sudo systemctl restart docker

部署K8s Node
复制程序文件

cd /tmp
for h in k8s-master k8s-node01 k8s-node02; do scp kubernetes/server/bin/kubelet kubernetes/server/bin/kube-proxy user@$h:~/kube; done
for h in k8s-master k8s-node01 k8s-node02; do scp flannel-${FLANNEL_VERSION}/flanneld user@$h:~/kube;done
for h in k8s-master k8s-node01 k8s-node02; do ssh -t user@$h 'sudo mkdir -p /opt/bin && sudo mv ~/kube/* /opt/bin/'; done

配置Flanned以及修改Docker服务

参见Master部分相关步骤： 配置Flanneld服务，启动Flanneld服务，修改Docker服务。注意修改iface的地址

配置kubelet服务

/lib/systemd/system/kubelet.service，注意修改IP地址

[Unit]
Description=Kubernetes Kubelet
After=docker.service
Requires=docker.service

[Service]
ExecStart=/opt/bin/kubelet \
  --hostname-override=172.12.24.37  \
  --api-servers=http://172.12.24.36:8080  \
  --logtostderr=true
Restart=on-failure
KillMode=process

[Install]
WantedBy=multi-user.target

启动服务
sudo systemctl daemon-reload
sudo systemctl enable kubelet
sudo systemctl start kubelet

配置kube-proxy服务
/lib/systemd/system/kube-proxy.service，注意修改IP地址
[Unit]
Description=Kubernetes Proxy
After=network.target

[Service]
ExecStart=/opt/bin/kube-proxy  \
  --hostname-override=172.12.24.37 \
  --master=http://172.12.24.36:8080 \
  --logtostderr=true
Restart=on-failure

[Install]
WantedBy=multi-user.target

启动服务
sudo systemctl daemon-reload
sudo systemctl enable kube-proxy
sudo systemctl start kube-proxy

部署附件组件
部署DNS

DNS_SERVER_IP="172.18.12.12"
DNS_DOMAIN="cluster.local"
DNS_REPLICAS=1
KUBE_APISERVER_URL=http://172.12.24.36:8080

cat <<EOF > skydns.yml
apiVersion: v1
kind: ReplicationController
metadata:
  name: kube-dns-v17.1
  namespace: kube-system
  labels:
    k8s-app: kube-dns
    version: v17.1
    kubernetes.io/cluster-service: "true"
spec:
  replicas: $DNS_REPLICAS
  selector:
    k8s-app: kube-dns
    version: v17.1
  template:
    metadata:
      labels:
        k8s-app: kube-dns
        version: v17.1
        kubernetes.io/cluster-service: "true"
    spec:
      containers:
      - name: kubedns
        image: gcr.io/google_containers/kubedns-amd64:1.5
        resources:
          # TODO: Set memory limits when we've profiled the Container for large
          # clusters, then set request = limit to keep this container in
          # guaranteed class. Currently, this container falls into the
          # "burstable" category so the kubelet doesn't backoff from restarting it.
          limits:
            cpu: 100m
            memory: 170Mi
          requests:
            cpu: 100m
            memory: 70Mi
        livenessProbe:
          httpGet:
            path: /healthz
            port: 8080
            scheme: HTTP
          initialDelaySeconds: 60
          timeoutSeconds: 5
          successThreshold: 1
          failureThreshold: 5
        readinessProbe:
          httpGet:
            path: /readiness
            port: 8081
            scheme: HTTP
          # we poll on pod startup for the Kubernetes master service and
          # only setup the /readiness HTTP server once that's available.
          initialDelaySeconds: 30
          timeoutSeconds: 5
        args:
        # command = "/kube-dns"
        - --domain=$DNS_DOMAIN.
        - --dns-port=10053
        - --kube-master-url=$KUBE_APISERVER_URL
        ports:
        - containerPort: 10053
          name: dns-local
          protocol: UDP
        - containerPort: 10053
          name: dns-tcp-local
          protocol: TCP
      - name: dnsmasq
        image: gcr.io/google_containers/kube-dnsmasq-amd64:1.3
        args:
        - --cache-size=1000
        - --no-resolv
        - --server=127.0.0.1#10053
        ports:
        - containerPort: 53
          name: dns
          protocol: UDP
        - containerPort: 53
          name: dns-tcp
          protocol: TCP
      - name: healthz
        image: gcr.io/google_containers/exechealthz-amd64:1.1
        resources:
          # keep request = limit to keep this container in guaranteed class
          limits:
            cpu: 10m
            memory: 50Mi
          requests:
            cpu: 10m
            # Note that this container shouldn't really need 50Mi of memory. The
            # limits are set higher than expected pending investigation on #29688.
            # The extra memory was stolen from the kubedns container to keep the
            # net memory requested by the pod constant.
            memory: 50Mi
        args:
        - -cmd=nslookup kubernetes.default.svc.$DNS_DOMAIN 127.0.0.1 >/dev/null
        - -port=8080
        - -quiet
        ports:
        - containerPort: 8080
          protocol: TCP
      dnsPolicy: Default  # Don't use cluster DNS.
---
apiVersion: v1
kind: Service
metadata:
  name: kube-dns
  namespace: kube-system
  labels:
    k8s-app: kube-dns
    kubernetes.io/cluster-service: "true"
    kubernetes.io/name: "KubeDNS"
spec:
  selector:
    k8s-app: kube-dns
  clusterIP: $DNS_SERVER_IP
  ports:
  - name: dns
    port: 53
    protocol: UDP
  - name: dns-tcp
    port: 53
    protocol: TCP
EOF

kubectl create -f skydns.yml

然后，修该各节点的kubelet.service,添加--cluster-dns=172.18.8.8以及--cluster-domain=cluster.local

部署Dashboard

echo <<'EOF' > kube-dashboard.yml
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
  labels:
    app: kubernetes-dashboard
    version: v1.1.0
  name: kubernetes-dashboard
  namespace: kube-system
spec:
  replicas: 1
  selector:
    matchLabels:
      app: kubernetes-dashboard
  template:
    metadata:
      labels:
        app: kubernetes-dashboard
    spec:
      containers:
      - name: kubernetes-dashboard
        image: gcr.io/google_containers/kubernetes-dashboard-amd64:v1.1.0
        imagePullPolicy: Always
        ports:
        - containerPort: 9090
          protocol: TCP
        args:
        - --apiserver-host=http://172.12.24.36:8080
        livenessProbe:
          httpGet:
            path: /
            port: 9090
          initialDelaySeconds: 30
          timeoutSeconds: 30
---
kind: Service
apiVersion: v1
metadata:
  labels:
    app: kubernetes-dashboard
  name: kubernetes-dashboard
  namespace: kube-system
spec:
  type: ClusterIP
  ports:
  - port: 80
    targetPort: 9090
  selector:
    app: kubernetes-dashboard
EOF

kubectl create -f kube-dashboard.yml