通过 Kubeadm 安装 K8S 与高可用
文章目录
!版权声明:本博客内容均为原创,每篇博文作为知识积累,写博不易,转载请注明出处。
环境介绍:
- CentOS: 7.6
- Docker: 18.06.1-ce
- Kubernetes: 1.13.4
- Kuberadm: 1.13.4
- Kuberlet: 1.13.4
- Kuberctl: 1.13.4
部署介绍:
创建高可用首先先有一个 Master 节点,然后再让其他服务器加入组成三个 Master 节点高可用,然后再讲工作节点 Node 加入。下面将描述每个节点要执行的步骤:
- Master01: 二、三、四、五、六、七、八、九、十一
- Master02、Master03: 二、三、五、六、四、九
- node01、node02: 二、五、六、九
集群架构:
一、kuberadm 简介
Kuberadm 作用
Kubeadm 是一个工具,它提供了 kubeadm init 以及 kubeadm join 这两个命令作为快速创建 kubernetes 集群的最佳实践。
kubeadm 通过执行必要的操作来启动和运行一个最小可用的集群。它被故意设计为只关心启动集群,而不是之前的节点准备工作。同样的,诸如安装各种各样值得拥有的插件,例如 Kubernetes Dashboard、监控解决方案以及特定云提供商的插件,这些都不在它负责的范围。
相反,我们期望由一个基于 kubeadm 从更高层设计的更加合适的工具来做这些事情;并且,理想情况下,使用 kubeadm 作为所有部署的基础将会使得创建一个符合期望的集群变得容易。
Kuberadm 功能
- kubeadm init: 启动一个 Kubernetes 主节点
- kubeadm join: 启动一个 Kubernetes 工作节点并且将其加入到集群
- kubeadm upgrade: 更新一个 Kubernetes 集群到新版本
- kubeadm config: 如果使用 v1.7.x 或者更低版本的 kubeadm 初始化集群,您需要对集群做一些配置以便使用 kubeadm upgrade 命令
- kubeadm token: 管理 kubeadm join 使用的令牌
- kubeadm reset: 还原 kubeadm init 或者 kubeadm join 对主机所做的任何更改
- kubeadm version: 打印 kubeadm 版本
- kubeadm alpha: 预览一组可用的新功能以便从社区搜集反馈
功能版本
Area | Maturity Level |
---|---|
Command line UX | GA |
Implementation | GA |
Config file API | beta |
CoreDNS | GA |
kubeadm alpha subcommands | alpha |
High availability | alpha |
DynamicKubeletConfig | alpha |
Self-hosting | alpha |
二、前期准备
1、虚拟机分配说明
地址 | 主机名 | 内存&CPU | 角色 |
---|---|---|---|
192.168.2.10 | --- | --- | vip |
192.168.2.11 | k8s-master-01 | 2C & 2G | master |
192.168.2.12 | k8s-master-02 | 2C & 2G | master |
192.168.2.13 | k8s-master-03 | 2C & 2G | master |
192.168.2.21 | k8s-node-01 | 2c & 4G | node |
192.168.2.22 | k8s-node-02 | 2c & 4G | node |
2、各个节点端口占用
- Master 节点
规则 | 方向 | 端口范围 | 作用 | 使用者 |
---|---|---|---|---|
TCP | Inbound | 6443* | Kubernetes API | server All |
TCP | Inbound | 2379-2380 | etcd server | client API kube-apiserver, etcd |
TCP | Inbound | 10250 | Kubelet API | Self, Control plane |
TCP | Inbound | 10251 | kube-scheduler | Self |
TCP | Inbound | 10252 | kube-controller-manager | Sel |
- node 节点
规则 | 方向 | 端口范围 | 作用 | 使用者 |
---|---|---|---|---|
TCP | Inbound | 10250 | Kubelet API | Self, Control plane |
TCP | Inbound | 30000-32767 | NodePort Services** | All |
3、基础环境设置
Kubernetes 需要一定的系统环境来保证正常运行,如各个节点时间同步,主机名称解析,关闭防火墙等等。
主机名称解析
修改hosts
分别进入不同服务器,进入 /etc/hosts 进行编辑
1vim /etc/hosts
加入下面内容:
1192.168.2.10 master.k8s.io k8s-vip
2192.168.2.11 master01.k8s.io k8s-master-01
3192.168.2.12 master02.k8s.io k8s-master-02
4192.168.2.13 master03.k8s.io k8s-master-03
5192.168.2.21 node01.k8s.io k8s-node-01
6192.168.2.22 node02.k8s.io k8s-node-02
修改hostname
分别进入不同的服务器,修改 hostname 名称
1# 修改 192.168.2.11 服务器
2hostnamectl set-hostname k8s-master-01
3# 修改 192.168.2.12 服务器
4hostnamectl set-hostname k8s-master-02
5# 修改 192.168.2.13 服务器
6hostnamectl set-hostname k8s-master-03
7
8# 修改 192.168.2.21 服务器
9hostnamectl set-hostname k8s-node-01
10# 修改 192.168.2.22 服务器
11hostnamectl set-hostname k8s-node-02
主机时间同步
将各个服务器的时间同步,并设置开机启动同步时间服务
1systemctl start chronyd.service
2systemctl enable chronyd.service
关闭防火墙服务
1systemctl stop firewalld
2systemctl disable firewalld
关闭并禁用SELinux
1# 关闭selinux
2setenforce 0
3
4# 编辑/etc/sysconfig selinux 文件,以彻底禁用 SELinux
5sed -i 's/^SELINUX=enforcing$/SELINUX=disabled/' /etc/selinux/config
6
7# 查看selinux状态
8getenforce
禁用 Swap 设备
Kubeadm 默认会预先检当前主机是否禁用了 Swap 设备,如果启动了 Swap 将导致安装不能正常进行,所以需要禁用所有的 Swap 设备。
1# 关闭当前已启用的所有 Swap 设备
2swapoff -a && sysctl -w vm.swappiness=0
1# 编辑 fstab 配置文件,注释掉标识为 Swap 设备的所有行
2vi /etc/fstab
设置系统参数
设置允许路由转发,不对 bridge 的数据进行处理,这里创建 /etc/sysctl.d/k8s.conf 文件,在里面加入对应配置。
1vim /etc/sysctl.d/k8s.conf
加入下面内容:
1net.ipv4.ip_forward = 1
2net.bridge.bridge-nf-call-ip6tables = 1
3net.bridge.bridge-nf-call-iptables = 1
挂载br_netfilter
1modprobe br_netfilter
使配置生效
1sysctl -p /etc/sysctl.d/k8s.conf
sysctl命令:用于运行时配置内核参数
查看是否生成相关文件
1ls /proc/sys/net/bridge
资源配置文件
/etc/security/limits.conf 是 Linux 资源使用配置文件,用来限制用户对系统资源的使用
1echo "* soft nofile 65536" >> /etc/security/limits.conf
2echo "* hard nofile 65536" >> /etc/security/limits.conf
3echo "* soft nproc 65536" >> /etc/security/limits.conf
4echo "* hard nproc 65536" >> /etc/security/limits.conf
5echo "* soft memlock unlimited" >> /etc/security/limits.conf
6echo "* hard memlock unlimited" >> /etc/security/limits.conf
安装依赖包以及相关工具
1yum install -y epel-release
1yum install -y yum-utils device-mapper-persistent-data lvm2 net-tools conntrack-tools wget vim ntpdate libseccomp libtool-ltdl
三、安装Keepalived
- keepalived 介绍: 是集群管理中保证集群高可用的一个服务软件,其功能类似于heartbeat,用来防止单点故障
- Keepalived 作用: 为haproxy提供vip(192.168.2.10)在三个haproxy实例之间提供主备,降低当其中一个haproxy失效的时对服务的影响。
1、yum安装Keepalived
1# 安装keepalived
2yum install -y keepalived
2、配置Keepalived
1cat <<EOF > /etc/keepalived/keepalived.conf
2! Configuration File for keepalived
3
4# 主要是配置故障发生时的通知对象以及机器标识。
5global_defs {
6 # 标识本节点的字条串,通常为 hostname,但不一定非得是 hostname。故障发生时,邮件通知会用到。
7 router_id LVS_k8s
8}
9
10# 用来做健康检查的,当时检查失败时会将 vrrp_instance 的 priority 减少相应的值。
11vrrp_script check_haproxy {
12 script "killall -0 haproxy" #根据进程名称检测进程是否存活
13 interval 3
14 weight -2
15 fall 10
16 rise 2
17}
18
19# rp_instance用来定义对外提供服务的 VIP 区域及其相关属性。
20vrrp_instance VI_1 {
21 state MASTER #当前节点为MASTER,其他两个节点设置为BACKUP
22 interface ens33 #改为自己的网卡
23 virtual_router_id 51
24 priority 250
25 advert_int 1
26 authentication {
27 auth_type PASS
28 auth_pass 35f18af7190d51c9f7f78f37300a0cbd
29 }
30 virtual_ipaddress {
31 192.168.2.10 #虚拟ip,即VIP
32 }
33 track_script {
34 check_haproxy
35 }
36
37}
38EOF
当前节点的配置中 state 配置为 MASTER,其它两个节点设置为 BACKUP
配置说明:
- virtual_ipaddress: vip
- track_script: 执行上面定义好的检测的script
- interface: 节点固有IP(非VIP)的网卡,用来发VRRP包。
- virtual_router_id: 取值在0-255之间,用来区分多个instance的VRRP组播
- advert_int: 发VRRP包的时间间隔,即多久进行一次master选举(可以认为是健康查检时间间隔)。
- authentication: 认证区域,认证类型有PASS和HA(IPSEC),推荐使用PASS(密码只识别前8位)。
- state: 可以是MASTER或BACKUP,不过当其他节点keepalived启动时会将priority比较大的节点选举为MASTER,因此该项其实没有实质用途。
- priority: 用来选举master的,要成为master,那么这个选项的值最好高于其他机器50个点,该项取值范围是1-255(在此范围之外会被识别成默认值100)。
3、启动Keepalived
1# 设置开机启动
2systemctl enable keepalived
3# 启动keepalived
4systemctl start keepalived
5# 查看启动状态
6systemctl status keepalived
4、查看网络状态
kepplived 配置中 state 为 MASTER 的节点启动后,查看网络状态,可以看到虚拟IP已经加入到绑定的网卡中
1ip address show ens33
当关掉当前节点的keeplived服务后将进行虚拟IP转移,将会推选state 为 BACKUP 的节点的某一节点为新的MASTER,可以在那台节点上查看网卡,将会查看到虚拟IP
四、安装haproxy
此处的haproxy为apiserver提供反向代理,haproxy将所有请求轮询转发到每个master节点上。相对于仅仅使用keepalived主备模式仅单个master节点承载流量,这种方式更加合理、健壮。
1、yum安装haproxy
1yum install -y haproxy
2、配置haproxy
1cat > /etc/haproxy/haproxy.cfg << EOF
2#---------------------------------------------------------------------
3# Global settings
4#---------------------------------------------------------------------
5global
6 # to have these messages end up in /var/log/haproxy.log you will
7 # need to:
8 # 1) configure syslog to accept network log events. This is done
9 # by adding the '-r' option to the SYSLOGD_OPTIONS in
10 # /etc/sysconfig/syslog
11 # 2) configure local2 events to go to the /var/log/haproxy.log
12 # file. A line like the following can be added to
13 # /etc/sysconfig/syslog
14 #
15 # local2.* /var/log/haproxy.log
16 #
17 log 127.0.0.1 local2
18
19 chroot /var/lib/haproxy
20 pidfile /var/run/haproxy.pid
21 maxconn 4000
22 user haproxy
23 group haproxy
24 daemon
25
26 # turn on stats unix socket
27 stats socket /var/lib/haproxy/stats
28#---------------------------------------------------------------------
29# common defaults that all the 'listen' and 'backend' sections will
30# use if not designated in their block
31#---------------------------------------------------------------------
32defaults
33 mode http
34 log global
35 option httplog
36 option dontlognull
37 option http-server-close
38 option forwardfor except 127.0.0.0/8
39 option redispatch
40 retries 3
41 timeout http-request 10s
42 timeout queue 1m
43 timeout connect 10s
44 timeout client 1m
45 timeout server 1m
46 timeout http-keep-alive 10s
47 timeout check 10s
48 maxconn 3000
49#---------------------------------------------------------------------
50# kubernetes apiserver frontend which proxys to the backends
51#---------------------------------------------------------------------
52frontend kubernetes-apiserver
53 mode tcp
54 bind *:16443
55 option tcplog
56 default_backend kubernetes-apiserver
57#---------------------------------------------------------------------
58# round robin balancing between the various backends
59#---------------------------------------------------------------------
60backend kubernetes-apiserver
61 mode tcp
62 balance roundrobin
63 server master01.k8s.io 192.168.2.11:6443 check
64 server master02.k8s.io 192.168.2.12:6443 check
65 server master03.k8s.io 192.168.2.13:6443 check
66#---------------------------------------------------------------------
67# collection haproxy statistics message
68#---------------------------------------------------------------------
69listen stats
70 bind *:1080
71 stats auth admin:awesomePassword
72 stats refresh 5s
73 stats realm HAProxy\ Statistics
74 stats uri /admin?stats
75EOF
haproxy配置在其他master节点上(192.168.2.12和192.168.2.13)相同
3、启动并检测haproxy
1# 设置开机启动
2systemctl enable haproxy
3# 开启haproxy
4systemctl start haproxy
5# 查看启动状态
6systemctl status haproxy
4、检测haproxy端口
1ss -lnt | grep -E "16443|1080"
显示:
五、安装Docker (所有节点)
1、移除之前安装过的Docker
1sudo yum remove docker \
2 docker-client \
3 docker-client-latest \
4 docker-common \
5 docker-latest \
6 docker-latest-logrotate \
7 docker-logrotate \
8 docker-selinux \
9 docker-engine-selinux \
10 docker-ce-cli \
11 docker-engine
查看还有没有存在的docker组件
1rpm -qa|grep docker
有则通过命令 yum -y remove XXX 来删除,比如:
1yum remove docker-ce-cli
2、配置docker的yum源
安装 Docker 可以从官方 yum 源下载对应应用,不过由于官方下载速度比较慢,所以推荐用阿里镜像源:
1sudo yum-config-manager --add-repo http://mirrors.aliyun.com/docker-ce/linux/centos/docker-ce.repo
3、安装Docker:
显示docker-ce所有可安装版本:
1yum list docker-ce --showduplicates | sort -r
安装指定docker版本
1sudo yum install docker-ce-18.06.1.ce-3.el7 -y
设置镜像存储目录
找到大点的挂载的目录进行存储
1# 修改docker配置
2vi /lib/systemd/system/docker.service
3
4#找到 ExecStart 这行,王后面加上存储目录,例如这里是 --graph /apps/docker
5ExecStart=/usr/bin/docker --graph /apps/docker
启动docker并设置docker开机启动
1systemctl enable docker
2systemctl start docker
4、确认一下 iptables
确认一下iptables filter表中FOWARD链的默认策略(pllicy)为ACCEPT。
1iptables -nvL
显示:
1Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
2 pkts bytes target prot opt in out source destination
3 0 0 DOCKER-USER all -- * * 0.0.0.0/0 0.0.0.0/0
4 0 0 DOCKER-ISOLATION-STAGE-1 all -- * * 0.0.0.0/0 0.0.0.0/0
5 0 0 ACCEPT all -- * docker0 0.0.0.0/0 0.0.0.0/0 ctstate RELATED,ESTABLISHED
6 0 0 DOCKER all -- * docker0 0.0.0.0/0 0.0.0.0/0
7 0 0 ACCEPT all -- docker0 !docker0 0.0.0.0/0 0.0.0.0/0
8 0 0 ACCEPT all -- docker0 docker0 0.0.0.0/0 0.0.0.0/0
Docker从1.13版本开始调整了默认的防火墙规则,禁用了iptables filter表中FOWARD链,这样会引起Kubernetes集群中跨Node的Pod无法通信。但这里通过安装docker 1806,发现默认策略又改回了ACCEPT,这个不知道是从哪个版本改回的,因为我们线上版本使用的1706还是需要手动调整这个策略的。
六、安装kubeadm、kubelet
-
需要在每台机器上都安装以下的软件包:
- kubeadm: 用来初始化集群的指令。
- kubelet: 在集群中的每个节点上用来启动 pod 和 container 等。
- kubectl: 用来与集群通信的命令行工具。
1、配置可用的国内yum源用于安装:
1cat <<EOF > /etc/yum.repos.d/kubernetes.repo
2[kubernetes]
3name=Kubernetes
4baseurl=https://mirrors.aliyun.com/kubernetes/yum/repos/kubernetes-el7-x86_64/
5enabled=1
6gpgcheck=0
7repo_gpgcheck=0
8gpgkey=https://mirrors.aliyun.com/kubernetes/yum/doc/yum-key.gpg https://mirrors.aliyun.com/kubernetes/yum/doc/rpm-package-key.gpg
9EOF
2、安装kubelet
查看kubelet版本列表
1yum list kubectl --showduplicates | sort -r
安装kubelet
1yum install -y kubectl-1.13.4-0
3、安装kubelet
查看kubelet版本列表
1yum list kubelet --showduplicates | sort -r
安装kubelet
1yum install -y kubelet-1.13.4-0
启动kubelet并设置开机启动
1systemctl enable kubelet
2systemctl start kubelet
检查状态
检查状态,发现是failed状态,正常,kubelet会10秒重启一次,等初始化master节点后即可正常
1systemctl status kubelet
4、安装kubeadm
查看kubeadm版本列表
1yum list kubeadm --showduplicates | sort -r
安装kubeadm
1yum install -y kubeadm-1.13.4-0
安装 kubeadm 时候会默认安装 kubectl ,所以不需要单独安装kubectl
4、重启服务器
为了防止发生某些未知错误,这里我们重启下服务器,方便进行后续操作
1reboot
七、初始化第一个kubernetes master节点
因为需要绑定虚拟IP,所以需要首先先查看虚拟IP启动这几台master机子哪台上
1ip address show ens33
显示:
1 ens33: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
2 link/ether 00:0c:29:7e:65:b3 brd ff:ff:ff:ff:ff:ff
3 inet 192.168.2.11/24 brd 192.168.2.255 scope global noprefixroute ens33
4 valid_lft forever preferred_lft forever
5 inet 192.168.2.10/32 scope global ens33
6 valid_lft forever preferred_lft forever
可以看到 10虚拟ip 和 11的ip 在一台机子上,所以初始化kubernetes第一个master要在master01机子上进行安装
1、创建kubeadm配置的yaml文件
1cat > kubeadm-config.yaml << EOF
2apiServer:
3 certSANs:
4 - k8s-master-01
5 - k8s-master-02
6 - k8s-master-03
7 - master.k8s.io
8 - 192.168.2.10
9 - 192.168.2.11
10 - 192.168.2.12
11 - 192.168.2.13
12 - 127.0.0.1
13 extraArgs:
14 authorization-mode: Node,RBAC
15 timeoutForControlPlane: 4m0s
16apiVersion: kubeadm.k8s.io/v1beta1
17certificatesDir: /etc/kubernetes/pki
18clusterName: kubernetes
19controlPlaneEndpoint: "master.k8s.io:16443"
20controllerManager: {}
21dns:
22 type: CoreDNS
23etcd:
24 local:
25 dataDir: /var/lib/etcd
26imageRepository: registry.aliyuncs.com/google_containers
27kind: ClusterConfiguration
28kubernetesVersion: v1.13.4
29networking:
30 dnsDomain: cluster.local
31 podSubnet: 10.20.0.0/16
32 serviceSubnet: 10.10.0.0/16
33scheduler: {}
34EOF
以下两个地方设置:
- certSANs: 虚拟ip地址(为了安全起见,把所有集群地址都加上)
- controlPlaneEndpoint: 虚拟IP:监控端口号
配置说明:
- imageRepository: registry.aliyuncs.com/google_containers (使用阿里云镜像仓库)
- podSubnet: 10.20.0.0/16 (pod地址池)
- serviceSubnet: 10.10.0.0/16
2、初始化第一个master节点
1kubeadm init --config kubeadm-config.yaml
日志:
1Your Kubernetes master has initialized successfully!
2
3To start using your cluster, you need to run the following as a regular user:
4
5 mkdir -p $HOME/.kube
6 sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
7 sudo chown $(id -u):$(id -g) $HOME/.kube/config
8
9You should now deploy a pod network to the cluster.
10Run "kubectl apply -f [podnetwork].yaml" with one of the options listed at:
11 https://kubernetes.io/docs/concepts/cluster-administration/addons/
12
13You can now join any number of machines by running the following on each node
14as root:
15
16 kubeadm join master.k8s.io:16443 --token dm3cw1.kw4hq84ie1376hji --discovery-token-ca-cert-hash sha256:f079b624773145ba714b56e177f52143f90f75a1dcebabda6538a49e224d4009
在此处看日志可以知道,通过
1kubeadm join master.k8s.io:16443 --token dm3cw1.kw4hq84ie1376hji --discovery-token-ca-cert-hash sha256:f079b624773145ba714b56e177f52143f90f75a1dcebabda6538a49e224d4009
来让节点加入集群
3、配置kubectl环境变量
配置环境变量
1mkdir -p $HOME/.kube
2sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
3sudo chown $(id -u):$(id -g) $HOME/.kube/config
4、查看组件状态
1kubectl get cs
显示:
1NAME STATUS MESSAGE ERROR
2controller-manager Healthy ok
3scheduler Healthy ok
4etcd-0 Healthy {"health": "true"}
查看pod状态
1kubectl get pods --namespace=kube-system
显示:
可以看到coredns没有启动,这是由于还没有配置网络插件,接下来配置下后再重新查看启动状态
八、安装网络插件
1、配置flannel插件的yaml文件
1cat > kube-flannel.yaml << EOF
2---
3kind: ClusterRole
4apiVersion: rbac.authorization.k8s.io/v1beta1
5metadata:
6 name: flannel
7rules:
8 - apiGroups:
9 - ""
10 resources:
11 - pods
12 verbs:
13 - get
14 - apiGroups:
15 - ""
16 resources:
17 - nodes
18 verbs:
19 - list
20 - watch
21 - apiGroups:
22 - ""
23 resources:
24 - nodes/status
25 verbs:
26 - patch
27---
28kind: ClusterRoleBinding
29apiVersion: rbac.authorization.k8s.io/v1beta1
30metadata:
31 name: flannel
32roleRef:
33 apiGroup: rbac.authorization.k8s.io
34 kind: ClusterRole
35 name: flannel
36subjects:
37- kind: ServiceAccount
38 name: flannel
39 namespace: kube-system
40---
41apiVersion: v1
42kind: ServiceAccount
43metadata:
44 name: flannel
45 namespace: kube-system
46---
47kind: ConfigMap
48apiVersion: v1
49metadata:
50 name: kube-flannel-cfg
51 namespace: kube-system
52 labels:
53 tier: node
54 app: flannel
55data:
56 cni-conf.json: |
57 {
58 "name": "cbr0",
59 "plugins": [
60 {
61 "type": "flannel",
62 "delegate": {
63 "hairpinMode": true,
64 "isDefaultGateway": true
65 }
66 },
67 {
68 "type": "portmap",
69 "capabilities": {
70 "portMappings": true
71 }
72 }
73 ]
74 }
75 net-conf.json: |
76 {
77 "Network": "10.20.0.0/16",
78 "Backend": {
79 "Type": "vxlan"
80 }
81 }
82---
83apiVersion: extensions/v1beta1
84kind: DaemonSet
85metadata:
86 name: kube-flannel-ds-amd64
87 namespace: kube-system
88 labels:
89 tier: node
90 app: flannel
91spec:
92 template:
93 metadata:
94 labels:
95 tier: node
96 app: flannel
97 spec:
98 hostNetwork: true
99 nodeSelector:
100 beta.kubernetes.io/arch: amd64
101 tolerations:
102 - operator: Exists
103 effect: NoSchedule
104 serviceAccountName: flannel
105 initContainers:
106 - name: install-cni
107 image: registry.cn-shenzhen.aliyuncs.com/cp_m/flannel:v0.10.0-amd64
108 command:
109 - cp
110 args:
111 - -f
112 - /etc/kube-flannel/cni-conf.json
113 - /etc/cni/net.d/10-flannel.conflist
114 volumeMounts:
115 - name: cni
116 mountPath: /etc/cni/net.d
117 - name: flannel-cfg
118 mountPath: /etc/kube-flannel/
119 containers:
120 - name: kube-flannel
121 image: registry.cn-shenzhen.aliyuncs.com/cp_m/flannel:v0.10.0-amd64
122 command:
123 - /opt/bin/flanneld
124 args:
125 - --ip-masq
126 - --kube-subnet-mgr
127 resources:
128 requests:
129 cpu: "100m"
130 memory: "50Mi"
131 limits:
132 cpu: "100m"
133 memory: "50Mi"
134 securityContext:
135 privileged: true
136 env:
137 - name: POD_NAME
138 valueFrom:
139 fieldRef:
140 fieldPath: metadata.name
141 - name: POD_NAMESPACE
142 valueFrom:
143 fieldRef:
144 fieldPath: metadata.namespace
145 volumeMounts:
146 - name: run
147 mountPath: /run
148 - name: flannel-cfg
149 mountPath: /etc/kube-flannel/
150 volumes:
151 - name: run
152 hostPath:
153 path: /run
154 - name: cni
155 hostPath:
156 path: /etc/cni/net.d
157 - name: flannel-cfg
158 configMap:
159 name: kube-flannel-cfg
160EOF
"Network": "10.20.0.0/16"要和kubeadm-config.yaml配置文件中podSubnet: 10.20.0.0/16相同
2、创建flanner相关role和pod
1kubectl apply -f kube-flannel.yaml
等待一会时间,再次查看各个pods的状态
1kubectl get pods --namespace=kube-system
显示:
可以看到coredns已经启动
九、加入集群
1、Master加入集群构成高可用
复制秘钥到各个节点
在master01 服务器上执行下面命令,将kubernetes相关文件复制到 master02、master03
如果其他节点为初始化第一个master节点,则将该节点的配置文件复制到其余两个主节点,例如master03为第一个master节点,则将它的k8s配置复制到master02和master01。
- 复制文件到 master02
1ssh root@master02.k8s.io mkdir -p /etc/kubernetes/pki/etcd
2scp /etc/kubernetes/admin.conf root@master02.k8s.io:/etc/kubernetes
3scp /etc/kubernetes/pki/{ca.*,sa.*,front-proxy-ca.*} root@master02.k8s.io:/etc/kubernetes/pki
4scp /etc/kubernetes/pki/etcd/ca.* root@master02.k8s.io:/etc/kubernetes/pki/etcd
- 复制文件到 master03
1ssh root@master03.k8s.io mkdir -p /etc/kubernetes/pki/etcd
2scp /etc/kubernetes/admin.conf root@master03.k8s.io:/etc/kubernetes
3scp /etc/kubernetes/pki/{ca.*,sa.*,front-proxy-ca.*} root@master03.k8s.io:/etc/kubernetes/pki
4scp /etc/kubernetes/pki/etcd/ca.* root@master03.k8s.io:/etc/kubernetes/pki/etcd
- master节点加入集群
master02 和 master03 服务器上都执行加入集群操作
1kubeadm join master.k8s.io:16443 --token dm3cw1.kw4hq84ie1376hji --discovery-token-ca-cert-hash sha256:f079b624773145ba714b56e177f52143f90f75a1dcebabda6538a49e224d4009 --experimental-control-plane
如果加入失败想重新尝试,请输入 kubeadm reset 命令清除之前的设置,重新执行从“复制秘钥”和“加入集群”这两步
显示安装过程:
1......
2This node has joined the cluster and a new control plane instance was created:
3
4* Certificate signing request was sent to apiserver and approval was received.
5* The Kubelet was informed of the new secure connection details.
6* Master label and taint were applied to the new node.
7* The Kubernetes control plane instances scaled up.
8* A new etcd member was added to the local/stacked etcd cluster.
9
10To start administering your cluster from this node, you need to run the following as a regular user:
11
12 mkdir -p $HOME/.kube
13 sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
14 sudo chown $(id -u):$(id -g) $HOME/.kube/config
15
16Run 'kubectl get nodes' to see this node join the cluster.
- 配置kubectl环境变量
1mkdir -p $HOME/.kube
2sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
3sudo chown $(id -u):$(id -g) $HOME/.kube/config
2、node节点加入集群
除了让master节点加入集群组成高可用外,slave节点也要加入集群中。
这里将k8s-node-01、k8s-node-02加入集群,进行工作
输入初始化k8s master时候提示的加入命令,如下:
1kubeadm join master.k8s.io:16443 --token dm3cw1.kw4hq84ie1376hji --discovery-token-ca-cert-hash sha256:f079b624773145ba714b56e177f52143f90f75a1dcebabda6538a49e224d4009
3、如果忘记加入集群的token和sha256 (如正常则跳过)
- 显示获取token列表
1kubeadm token list
默认情况下 Token 过期是时间是24小时,如果 Token 过期以后,可以输入以下命令,生成新的 Token
1kubeadm token create
- 获取ca证书sha256编码hash值
1openssl x509 -pubkey -in /etc/kubernetes/pki/ca.crt | openssl rsa -pubin -outform der 2>/dev/null | openssl dgst -sha256 -hex | sed 's/^.* //'
- 拼接命令
1kubeadm join master.k8s.io:16443 --token 882ik4.9ib2kb0eftvuhb58 --discovery-token-ca-cert-hash sha256:0b1a836894d930c8558b350feeac8210c85c9d35b6d91fde202b870f3244016a
如果是master加入,请在最后面加上 --experimental-control-plane 这个参数
4、查看各个节点加入集群情况
1kubectl get nodes -o wide
显示:
1NAME STATUS ROLES AGE VERSION INTERNAL-IP EXTERNAL-IP OS-IMAGE KERNEL-VERSION CONTAINER-RUNTIME
2k8s-master-01 Ready master 12m v1.13.4 192.168.2.11 <none> CentOS Linux 7 (Core) 3.10.0-957.1.3.el7.x86_64 docker://18.6.1
3k8s-master-02 Ready master 10m v1.13.4 192.168.2.12 <none> CentOS Linux 7 (Core) 3.10.0-957.1.3.el7.x86_64 docker://18.6.1
4k8s-master-03 Ready master 38m v1.13.4 192.168.2.13 <none> CentOS Linux 7 (Core) 3.10.0-957.1.3.el7.x86_64 docker://18.6.1
5k8s-node-01 Ready <none> 68s v1.13.4 192.168.2.21 <none> CentOS Linux 7 (Core) 3.10.0-957.1.3.el7.x86_64 docker://18.6.1
6k8s-node-02 Ready <none> 61s v1.13.4 192.168.2.22 <none> CentOS Linux 7 (Core) 3.10.0-957.1.3.el7.x86_64 docker://18.6.1
十、从集群中删除 Node
- Master节点:
1kubectl drain <node name> --delete-local-data --force --ignore-daemonsets
2kubectl delete node <node name>
- slave节点
1kubeadm reset
十一、配置dashboard
这个在一个服务器上部署,其他服务器会复制这个部署的pod,所以这里在master01服务器上部署 dashboard
1、创建 dashboard.yaml 并启动
1# ------------------- Dashboard Secret ------------------- #
2apiVersion: v1
3kind: Secret
4metadata:
5 labels:
6 k8s-app: kubernetes-dashboard
7 name: kubernetes-dashboard-certs
8 namespace: kube-system
9type: Opaque
10
11---
12# ------------------- Dashboard Service Account ------------------- #
13
14apiVersion: v1
15kind: ServiceAccount
16metadata:
17 labels:
18 k8s-app: kubernetes-dashboard
19 name: kubernetes-dashboard
20 namespace: kube-system
21
22---
23# ------------------- Dashboard Role & Role Binding ------------------- #
24
25kind: Role
26apiVersion: rbac.authorization.k8s.io/v1
27metadata:
28 name: kubernetes-dashboard-minimal
29 namespace: kube-system
30rules:
31 # Allow Dashboard to create 'kubernetes-dashboard-key-holder' secret.
32- apiGroups: [""]
33 resources: ["secrets"]
34 verbs: ["create"]
35 # Allow Dashboard to create 'kubernetes-dashboard-settings' config map.
36- apiGroups: [""]
37 resources: ["configmaps"]
38 verbs: ["create"]
39 # Allow Dashboard to get, update and delete Dashboard exclusive secrets.
40- apiGroups: [""]
41 resources: ["secrets"]
42 resourceNames: ["kubernetes-dashboard-key-holder", "kubernetes-dashboard-certs"]
43 verbs: ["get", "update", "delete"]
44 # Allow Dashboard to get and update 'kubernetes-dashboard-settings' config map.
45- apiGroups: [""]
46 resources: ["configmaps"]
47 resourceNames: ["kubernetes-dashboard-settings"]
48 verbs: ["get", "update"]
49 # Allow Dashboard to get metrics from heapster.
50- apiGroups: [""]
51 resources: ["services"]
52 resourceNames: ["heapster"]
53 verbs: ["proxy"]
54- apiGroups: [""]
55 resources: ["services/proxy"]
56 resourceNames: ["heapster", "http:heapster:", "https:heapster:"]
57 verbs: ["get"]
58
59---
60
61apiVersion: rbac.authorization.k8s.io/v1
62kind: RoleBinding
63metadata:
64 name: kubernetes-dashboard-minimal
65 namespace: kube-system
66roleRef:
67 apiGroup: rbac.authorization.k8s.io
68 kind: Role
69 name: kubernetes-dashboard-minimal
70subjects:
71- kind: ServiceAccount
72 name: kubernetes-dashboard
73 namespace: kube-system
74
75---
76
77# ------------------- Dashboard Deployment ------------------- #
78# 1.修改了镜像仓库位置,编辑成自己的镜像仓库
79# 2.变更了镜像拉去策略imagePullPolicy: IfNotPresent
80kind: Deployment
81apiVersion: apps/v1
82metadata:
83 labels:
84 k8s-app: kubernetes-dashboard
85 name: kubernetes-dashboard
86 namespace: kube-system
87spec:
88 replicas: 1
89 revisionHistoryLimit: 10
90 selector:
91 matchLabels:
92 k8s-app: kubernetes-dashboard
93 template:
94 metadata:
95 labels:
96 k8s-app: kubernetes-dashboard
97 spec:
98 containers:
99 - name: kubernetes-dashboard
100 image: registry.cn-hangzhou.aliyuncs.com/google_containers/kubernetes-dashboard-amd64:v1.10.1
101 imagePullPolicy: IfNotPresent
102 ports:
103 - containerPort: 8443
104 protocol: TCP
105 args:
106 - --auto-generate-certificates
107 # Uncomment the following line to manually specify Kubernetes API server Host
108 # If not specified, Dashboard will attempt to auto discover the API server and connect
109 # to it. Uncomment only if the default does not work.
110 # - --apiserver-host=http://my-address:port
111 volumeMounts:
112 - name: kubernetes-dashboard-certs
113 mountPath: /certs
114 # Create on-disk volume to store exec logs
115 - mountPath: /tmp
116 name: tmp-volume
117 livenessProbe:
118 httpGet:
119 scheme: HTTPS
120 path: /
121 port: 8443
122 initialDelaySeconds: 30
123 timeoutSeconds: 30
124 volumes:
125 - name: kubernetes-dashboard-certs
126 secret:
127 secretName: kubernetes-dashboard-certs
128 - name: tmp-volume
129 emptyDir: {}
130 serviceAccountName: kubernetes-dashboard
131 # Comment the following tolerations if Dashboard must not be deployed on master
132 tolerations:
133 - key: node-role.kubernetes.io/master
134 effect: NoSchedule
135---
136# ------------------- Dashboard Service ------------------- #
137# 增加了nodePort,使得能够访问,改变默认的type类型ClusterIP,变为NodePort
138# 如果不配置的话默认只能集群内访问
139kind: Service
140apiVersion: v1
141metadata:
142 labels:
143 k8s-app: kubernetes-dashboard
144 name: kubernetes-dashboard
145 namespace: kube-system
146spec:
147 type: NodePort
148 ports:
149 - port: 443
150 targetPort: 8443
151 nodePort: 30001
152 selector:
153 k8s-app: kubernetes-dashboard
运行 dashboard
1kubectl create -f kubernetes-dashboard.yaml
2、Dashboard 创建 ServiceAccount 并绑定 Admin 角色
1kind: ClusterRoleBinding
2apiVersion: rbac.authorization.k8s.io/v1beta1
3metadata:
4 name: admin
5 annotations:
6 rbac.authorization.kubernetes.io/autoupdate: "true"
7roleRef:
8 kind: ClusterRole
9 name: cluster-admin
10 apiGroup: rbac.authorization.k8s.io
11subjects:
12- kind: ServiceAccount
13 name: admin
14 namespace: kube-system
15---
16apiVersion: v1
17kind: ServiceAccount
18metadata:
19 name: admin
20 namespace: kube-system
21 labels:
22 kubernetes.io/cluster-service: "true"
23 addonmanager.kubernetes.io/mode: Reconcile
运行dashboard的用户和角色绑定
1kubectl create -f dashboard-user-role.yaml
获取登陆token
1kubectl describe secret/$(kubectl get secret -n kube-system |grep admin|awk '{print $1}') -n kube-system
显示:
1[root@k8s-master-01 local]# kubectl describe secret/$(kubectl get secret -nkube-system |grep admin|awk '{print $1}') -nkube-system
2Name: admin-token-2mfdz
3Namespace: kube-system
4Labels: <none>
5Annotations: kubernetes.io/service-account.name: admin
6 kubernetes.io/service-account.uid: 74efd994-38d8-11e9-8740-000c299624e4
7
8Type: kubernetes.io/service-account-token
9
10Data
11====
12ca.crt: 1025 bytes
13namespace: 11 bytes
token:
eyJhbGciOiJSUzI1NiIsImtpZCI6IiJ9.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJrdWJlLXN5c3RlbSIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VjcmV0Lm5hbWUiOiJhZG1pbi10b2tlbi1qdjd4ayIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50Lm5hbWUiOiJhZG1pbiIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50LnVpZCI6ImM4ZTMxYzk0LTQ2MWEtMTFlOS1iY2M5LTAwMGMyOTEzYzUxZCIsInN1YiI6InN5c3RlbTpzZXJ2aWNlYWNjb3VudDprdWJlLXN5c3RlbTphZG1pbiJ9.TNw1iFEsZmJsVG4cki8iLtEoiY1pjpnOYm8ZIFjctpBdTOw6kUMvv2b2B2BJ_5rFle31gqGAZBIRyYj9LPAs06qT5uVP_l9o7IyFX4HToBF3veiun4e71822eQRUsgqiPh5uSjKXEkf9yGq9ujiCdtzFxnp3Pnpeuge73syuwd7J6F0-dJAp3b48MLZ1JJwEo6CTCMhm9buysycUYTbT_mUDQMNrHVH0868CdN_H8azA4PdLLLrFfTiVgoGu4c3sG5rgh9kKFqZA6dzV0Kq10W5JJwJRM1808ybLHyV9jfKN8N2_lZ7ehE6PbPU0cV-PyP74iA-HrzFW1yVwSLPVYA
3、运行dashboard并登陆
输入地址:https://192.168.2.10:30001 进入 dashboard 界面
这里输入上面获取的 token 进入 dashboard
问题
1、Master不会参与负载工作
Master不会参与负载工作,如何让其参加,这里需要了解traint
查看traint
1# 查看全部节点是否能被安排工作
2kubectl describe nodes | grep -E '(Roles|Taints)'
删除traint
1# 所有node都可以调度
2kubectl taint nodes --all node-role.kubernetes.io/master-
3# 指定node可以调度
4kubectl taint nodes k8s-master-01 node-role.kubernetes.io/master-
2、重新加入集群
有时候节点出现问题需要重新加入集群,加入前需要清除一些设置,不然可能出现某些错误,比如
1network is not ready: [runtime network not ready: NetworkReady=false reason:NetworkPluginNotReady message:docker: network plugin is not ready: cni config uninitialized]
2Back-off restarting failed container
按下面步骤执行,再执行加入集群命令即可
1#重置kubernetes服务,重置网络。删除网络配置,link
2kubeadm reset
3
4#停止docker
5systemctl stop docker
6
7#重置cni
8rm -rf /var/lib/cni/
9rm -rf /var/lib/kubelet/*
10rm -rf /etc/cni/
11ifconfig cni0 down
12ifconfig flannel.1 down
13ifconfig docker0 down
14ip link delete cni0
15ip link delete flannel.1
16
17#重启docker
18systemctl start docker
再次加入集群
1kubeadm join cluster.kube.com:16443 --token gaeyou.k2650x660c8eb98c --discovery-token-ca-cert-hash sha256:daf4c2e0264422baa7076a2587f9224a5bd9c5667307927b0238743799dfb362
!版权声明:本博客内容均为原创,每篇博文作为知识积累,写博不易,转载请注明出处。