云主机搭建Kubernetes 1.10集群
一、基础环境 云主机 下载软件包 将所有软件下载至/data目录 # 链接：https://pan.baidu.com/s/13DlR1akNBCjib5VFaIjGTQ 密码：1l69 # 链接：https://pan.baidu.com/s/1V6Uuj6a08mEq-mRLaI1xgw 密码：6gap master到node做免密认证 ssh-keygen ssh-copy-id root@192.168.1.237 ssh-copy-id root@192.168.1.100 ssh-copy-id root@192.168.1.188 设定主机名与host文件 # 分别设定node与master的主机名 hostnamectl set-hostname master exec bash # 同步所有主机的hosts文件 vim /etc/hosts 192.168.1.78 master localhost 192.168.1.237 node1 192.168.1.100 node2 192.168.1.188 node3 解决DNS解析localhost 此云主机的DNS解析localhost会解析到一个鬼地址，这是个大坑。kubeadm初始化是会用到localhost。如果你的主机能解析到自己的IP，那么这步可以跳过。如果不能则需要自己搭建一个DNS，将localhost解析到自己。 # 1.检测 [root@node2 ~]# nslookup localhost Server: 118.118.118.9 Address: 118.118.118.9#53 Non-authoritative answer: Name: localhost.openstacklocal Address: 183.136.168.91 # 2.搭建DNS yum -y install dnsmasq cp /etc/resolv.conf{,.bak} rm -rf /etc/resolv.conf echo -e "nameserver 127.0.0.1\nnameserver $(hostname -i)" >> /etc/resolv.conf chmod 444 /etc/resolv.conf chattr +i /etc/resolv.conf echo -e "server=8.8.8.8\nserver=8.8.4.4" > /etc/dnsmasq.conf echo -e "$(hostname -i)\tlocalhost.$(hostname -d)" >> /etc/hosts service dnsmasq restart # 3.再次检测 [root@master ~]# nslookup localhost Server: 127.0.0.1 Address: 127.0.0.1#53 Name: localhost Address: 192.168.1.78 同步系统时间 ntpdate 0.centos.pool.ntp.org 关闭防火墙 iptables -F systemctl stop firewalld systemctl disable firewalld 关闭SELinux & 关闭swap swapoff -a sed -i 's/.*swap.*/#&/' /etc/fstab setenforce 0 确认时区 timedatectl set-timezone Asia/Shanghai systemctl restart chronyd.service 修改系统参数 cat <<EOF > /etc/sysctl.d/k8s.conf net.bridge.bridge-nf-call-ip6tables = 1 net.bridge.bridge-nf-call-iptables = 1 EOF sysctl --system 安装docker tar -xvf docker-packages.tar cd docker-packages yum -y install local *.rpm systemctl start docker && systemctl enable docker 配置镜像加速器 vim /etc/docker/daemon.json { "registry-mirrors": ["https://lw9sjwma.mirror.aliyuncs.com"] } systemctl daemon-reload systemctl restart docker 配置k8s的yum源 vim /etc/yum.repos.d/k8s.repo [k8s] name=k8s baseurl=https://mirrors.aliyun.com/kubernetes/yum/repos/kubernetes-el7-x86_64/ gpgcheck=0 获取kube软件包 cd kube-packages-1.10.1 # 软件包在网盘中下载 tar -xvf kube-packages-1.10.1.tar cd kube-packages-1.10.1 yum -y install local *.rpm systemctl start kubelet && systemctl enable kubelet 统一k8s与docker的驱动 # 1.查看docker驱动 docker info | Cgroup Driver Cgroup Driver: cgroupfs # 修改k8s配置文件与docker保持一致 sed -i "s/cgroup-driver=systemd/cgroup-driver=cgroupfs/g" /etc/systemd/system/kubelet.service.d/10-kubeadm.conf 导入基础镜像 cd /data docker load -i k8s-images-1.10.tar.gz 二、初始化master节点 # 初始化master 指定的版本要与kubeadm版本一致 # kubeadm只给定了最少选项，集群名称等等都没有指定，kubeadm init [root@master ~]# kubeadm init --kubernetes-version=v1.10.1 --pod-network-cidr=10.244.0.0/16 # 初始化完成后得到如下信息 Your Kubernetes master has initialized successfully! To start using your cluster, you need to run the following as a regular user: mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/config You should now deploy a pod network to the cluster. Run "kubectl apply -f [podnetwork].yaml" with one of the options listed at: https://kubernetes.io/docs/concepts/cluster-administration/addons/ You can now join any number of machines by running the following on each node as root: kubeadm join 192.168.1.78:6443 --token qabol0.c2gq0uyfxvpqr8bu --discovery-token-ca-cert-hash sha256:2237ec7b8efd5a8f68adcb04900a0b17b9df2a78675a7d62b4aef644a7f62c05 # kubeadm join 是node节点加入集群的命令，注意token的有效期 若要通过普通用户使用群集，需要执行以下步骤 mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/config 基本命令 # 查看pods kubectl get pods # 查看系统pods [root@master ~]# kubectl get pods -n kube-system NAME READY STATUS RESTARTS AGE etcd-master 0/1 Pending 0 1s kube-apiserver-master 0/1 Pending 0 1s kube-controller-manager-master 0/1 Pending 0 1s kube-dns-86f4d74b45-d42zm 0/3 Pending 0 8h kube-proxy-884h6 1/1 NodeLost 0 8h kube-scheduler-master 0/1 Pending 0 1s # 查看集群各组件状态信息 [root@master ~]# kubectl get componentstatuses NAME STATUS MESSAGE ERROR scheduler Healthy ok controller-manager Healthy ok etcd-0 Healthy {"health": "true"} You have new mail in /var/spool/mail/root 三、node加入集群 # 确保node节点cgroup驱动保持一致 sed -i "s/cgroup-driver=systemd/cgroup-driver=cgroupfs/g" /etc/systemd/system/kubelet.service.d/10-kubeadm.conf # 命令来自集群初始化之后额显示中 kubeadm join 192.168.1.78:6443 --token v0866r.u7kvg5js1ah2u1bi --discovery-token-ca-cert-hash sha256:7b36794f4fa5121f6a5e309d0e312ded72997a88236a93ec7da3520e5aaccf0e # master节点查看nodes信息 [root@master data]# kubectl get nodes NAME STATUS ROLES AGE VERSION master NotReady master 57m v1.10.1 node1 NotReady <none> 27m v1.10.1 node2 NotReady <none> 11s v1.10.1 node3 NotReady <none> 4s v1.10.1 You have new mail in /var/spool/mail/root 四、部署网络 部署 flannel官网 flannel下载时不用科学上网,flannel的yml文件会自动去quay.io网站中下载镜像。 # 1.1使用软件包中的flannel,并指pod映射到哪个主机的网卡上面。 vim kube-flannel.yml command: [ "/opt/bin/flanneld", "--ip-masq", "--kube-subnet-mgr","-iface=eth0" ] # 以下要按顺序创建，先创建rbac，之前没有穿件rbac导致pod正常创建，但是pin不同 kubectl apply -f kube-flannel-rbac.yml kubectl apply -f kube-flannel.yml # 后，节点的状态会变为ready [root@master1 kubernetes1.10]# kubectl get node NAME STATUS ROLES AGE VERSION master Ready master 57m v1.10.1 node1 Ready <none> 27m v1.10.1 node2 Ready <none> 11s v1.10.1 node3 Ready <none> 4s v1.10.1 # 2.从官网下载最新的flannel，k8s1.7+ 直接执行以下命令即可 kubectl apply -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml flannel配置文件修改 kube-flannel.yml中指定使用的网段 "Network": "10.244.0.0/16" 默认使用16位掩码，则在各node中都分配一个10.244.0.0/8的网络 五、部署dashboard kubectl apply -f kubernetes-dashboard-http.yam kubectl apply -f admin-role.yaml kubectl apply -f kubernetes-dashboard-admin.rbac.yaml 命令行常用命令 # 查看pod信息，默认显示default名称空间下的pod [root@master ~]# kubectl get pods No resources found. # 指定名称空间写pod [root@master ~]# kubectl get pods -n kube-system NAME READY STATUS RESTARTS AGE etcd-master 1/1 Running 0 3h kube-apiserver-master 1/1 Running 0 3h kube-controller-manager-master 1/1 Running 0 3h kube-dns-86f4d74b45-bzbvc 3/3 Running 0 3h kube-flannel-ds-5ghhj 1/1 Running 0 2h kube-flannel-ds-ht4xd 1/1 Running 0 3h kube-flannel-ds-kbm5g 1/1 Running 0 3h kube-flannel-ds-mlj4r 1/1 Running 0 2h kube-proxy-9xxnd 1/1 Running 0 3h kube-proxy-n9w5x 1/1 Running 0 3h kube-proxy-nkn8c 1/1 Running 0 2h kube-proxy-shd6l 1/1 Running 0 2h kube-scheduler-master 1/1 Running 0 3h kubernetes-dashboard-5c469b58b8-rjfx6 1/1 Running 0 1h # 显示更详细的pod信息，此时各pod中都运行了一个kube-proxy和flannel容器 -o wide 显示更详细的信息，报错node节点iP、主机名 [root@master ~]# kubectl get pods -n kube-system -o wide NAME READY STATUS RESTARTS AGE IP NODE etcd-master 1/1 Running 0 3h 192.168.1.78 master kube-apiserver-master 1/1 Running 0 3h 192.168.1.78 master kube-controller-manager-master 1/1 Running 0 3h 192.168.1.78 master kube-dns-86f4d74b45-bzbvc 3/3 Running 0 3h 10.244.0.2 master kube-flannel-ds-5ghhj 1/1 Running 0 2h 192.168.1.188 node3 kube-flannel-ds-ht4xd 1/1 Running 0 3h 192.168.1.78 master kube-flannel-ds-kbm5g 1/1 Running 0 3h 192.168.1.237 node1 kube-flannel-ds-mlj4r 1/1 Running 0 2h 192.168.1.100 node2 kube-proxy-9xxnd 1/1 Running 0 3h 192.168.1.237 node1 kube-proxy-n9w5x 1/1 Running 0 3h 192.168.1.78 master kube-proxy-nkn8c 1/1 Running 0 2h 192.168.1.100 node2 kube-proxy-shd6l 1/1 Running 0 2h 192.168.1.188 node3 kube-scheduler-master 1/1 Running 0 3h 192.168.1.78 master kubernetes-dashboard-5c469b58b8-rjfx6 1/1 Running 0 1h 10.244.0.3 master 六、kubeadm清空配置 # 清空kubectl kubeadm reset # 清空网络信息 ip link del cni0 ip link del flannel.1 七、踩过的那些坑 确保master与node的DNS解析localhost能解析到自己的IP node加入master确保token不过期 node确保kubelet正常启动并运行 flannel网络要先创建kube-flannel-rbac.ymal再创建 kube-flannel.yml 八、token过期的解决办法 # 1.查看已经存在的token kubeadm token list # 2.创建token kubeadm token create # 3.查看ca证书的sha256编码 openssl x509 -pubkey -in /etc/kubernetes/pki/ca.crt | openssl rsa -pubin -outform der 2>/dev/null | openssl dgst -sha256 -hex | sed 's/^.* //' # 4.node使用新的token加入集群 kubeadm join --token acb123 --discovery-token-ca-cert-hash sha256:efg456 172.16.6.79:6443 --skip-preflight-checks # abc123 新创建的Token # efg456 证书的sha256编码 # IP+Port Master的IP+Port 感谢 无痴迷，不成功 discsthnew
