原生K8s环境下 TiDB Operator实战

作者：贾世闻

原文来源： https://tidb.net/blog/00e04215

凑个热闹，k8s+tidb operator，便于更好的了解结构和操作细节。 k8s和tidb都是目前开源社区中活跃的开源产品，tidb-operator项目是一个在k8s上编排管理tidb集群的项目。本文详细记录了部署k8s及install tidb-operator的详细实施过程，希望能对刚入坑的同学有所帮助

环境

• ubuntu 16.04
• k8s 1.14.1

kubespray安装k8s

• 参考文档
• [国内镜像方法] https://segmentfault.com/a/1190000015186299
• https://jicki.me/kubernetes/docker/2018/12/21/k8s-1.13.1-kubespray/

配置免密登录

yum -y install expect

• vi /tmp/autocopy.exp

#!/usr/bin/expect
 
set timeout
set user_hostname [lindex $argv ]
set password [lindex $argv ]
spawn ssh-copy-id $user_hostname
    expect {
        "(yes/no)?"
        {
            send "yes
"
            expect "*assword:" { send "xxxxxx
"}
        }
        "*assword:"
        {
            send "xxxxxx
"
        }
    }
expect eof

ssh-keyscan addedip  >> ~/.ssh/known_hosts

ssh-keygen -t rsa -P ''

for i in 10.0.0.{31,32,33,40,10,20，50}; do  ssh-keyscan $i  >> ~/.ssh/known_hosts ; done

/tmp/autocopy.exp root@addeip
ssh-copy-id addedip

/tmp/autocopy.exp [email protected]
/tmp/autocopy.exp [email protected]
/tmp/autocopy.exp [email protected]
/tmp/autocopy.exp [email protected]
/tmp/autocopy.exp [email protected]
/tmp/autocopy.exp [email protected]
/tmp/autocopy.exp [email protected]

配置kubespray

pip install -r requirements.txt
cp -rfp inventory/sample inventory/mycluster

inventory/mycluster/inventory.ini

# ## Configure 'ip' variable to bind kubernetes services on a
# ## different ip than the default iface
# ## We should set etcd_member_name for etcd cluster. The node that is not a etcd member do not need to set the value, or can set the empty string value.
[all]
# node1 ansible_host=95.54.0.12  # ip=10.3.0.1 etcd_member_name=etcd1
# node2 ansible_host=95.54.0.13  # ip=10.3.0.2 etcd_member_name=etcd2
# node3 ansible_host=95.54.0.14  # ip=10.3.0.3 etcd_member_name=etcd3
# node4 ansible_host=95.54.0.15  # ip=10.3.0.4 etcd_member_name=etcd4
# node5 ansible_host=95.54.0.16  # ip=10.3.0.5 etcd_member_name=etcd5
# node6 ansible_host=95.54.0.17  # ip=10.3.0.6 etcd_member_name=etcd6
etcd1 ansible_host=10.0.0.31 etcd_member_name=etcd1
etcd2 ansible_host=10.0.0.32 etcd_member_name=etcd2
etcd3 ansible_host=10.0.0.33 etcd_member_name=etcd3
master1 ansible_host=10.0.0.40
node1 ansible_host=10.0.0.10
node2 ansible_host=10.0.0.20
node3 ansible_host=10.0.0.50

# ## configure a bastion host if your nodes are not directly reachable
# bastion ansible_host=x.x.x.x ansible_user=some_user

[kube-master]
# node1
# node2
master1
[etcd]
# node1
# node2
# node3
etcd1
etcd2
etcd3

[kube-node]
# node2
# node3
# node4
# node5
# node6
node1
node2
node3

[k8s-cluster:children]
kube-master
kube-node

节点所需镜像的文件

由于某些镜像国内无法访问需要现将镜像通过代理下载到本地然后上传到本地镜像仓库或dockerhub，同时修改配置文件，个别组件存放位置 https://storage.googleapis.com，需要新建nginx服务器分发文件

• 建立nginx服务器

  • 安装docker及docker-compose

    apt-get install 
    apt-transport-https 
    ca-certificates 
    curl 
    gnupg-agent 
    software-properties-common
    
    curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
    
    add-apt-repository 
    "deb [arch=amd64] https://download.docker.com/linux/ubuntu 
    $(lsb_release -cs) 
    stable"
    
    apt-get update
    
    apt-get install docker-ce docker-ce-cli containerd.io
    
    
    sudo curl -L "https://github.com/docker/compose/releases/download/1.24.0/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
    chmod +x /usr/local/bin/docker-compose
    

  • 新建nginx docker-compose.yml

    mkdir ~/distribution
    vi ~/distribution/docker-compose.yml
    

    • ~/distribution/docker-compose.yml

    #  distribute
    version: '2'
    services:    
        distribute:
            image: nginx:1.15.12
            volumes:
                - ./conf.d:/etc/nginx/conf.d
                - ./distributedfiles:/usr/share/nginx/html
            network_mode: "host"
            container_name: nginx_distribute 
    

    • 创建文件目录及nginx配置文件目录

    mkdir ~/distribution/distributedfiles
    mkdir ~/distribution/
    mkdir ~/distribution/conf.d
    vi ~/distribution/conf.d/open_distribute.conf
    

    • ~/distribution/conf.d/open_distribute.conf

    #open_distribute.conf
    
    server {
    	#server_name distribute.search.leju.com;
            listen 8888;
    
    	root /usr/share/nginx/html;
    
    	add_header Access-Control-Allow-Origin *;  
    	add_header Access-Control-Allow-Headers X-Requested-With;  
    	add_header Access-Control-Allow-Methods GET,POST,OPTIONS;  
    
    	location / {
    	#	 index index.html;
                    autoindex on;        
    	}
    	expires off;
    	location ~ .*.(gif|jpg|jpeg|png|bmp|swf|eot|ttf|woff|woff2|svg)$ {
    		expires -1;
    	}
    
    	location ~ .*.(js|css)?$ {
    		expires -1 ;
    	}
    } # end of public static files domain : [ distribute.search.leju.com ]
    

    • 启动

    docker-compose up -d
    

    • 下载并上传所需文件 具体版本号参考roles/download/defaults/main.yml文件中kubeadm_version、kube_version、image_arch参数

    wget https://storage.googleapis.com/kubernetes-release/release/v1.14.1/bin/linux/amd64/kubeadm
    
    scp /tmp/kubeadm  10.0.0.60:/root/distribution/distributedfiles
    
    wget https://storage.googleapis.com/kubernetes-release/release/v1.14.1/bin/linux/amd64/hyperkube
    

• 需要下载并上传到私有仓库的镜像

  docker pull k8s.gcr.io/cluster-proportional-autoscaler-amd64:1.4.0
  docker tag k8s.gcr.io/cluster-proportional-autoscaler-amd64:1.4.0 jiashiwen/cluster-proportional-autoscaler-amd64:1.4.0
  docker push jiashiwen/cluster-proportional-autoscaler-amd64:1.4.0
  
  docker pull k8s.gcr.io/k8s-dns-node-cache:1.15.1
  docker tag k8s.gcr.io/k8s-dns-node-cache:1.15.1 jiashiwen/k8s-dns-node-cache:1.15.1
  docker push jiashiwen/k8s-dns-node-cache:1.15.1
  
  docker pull gcr.io/google_containers/pause-amd64:3.1
  docker tag gcr.io/google_containers/pause-amd64:3.1 jiashiwen/pause-amd64:3.1
  docker push jiashiwen/pause-amd64:3.1
  
  docker pull gcr.io/google_containers/kubernetes-dashboard-amd64:v1.10.1
  docker tag gcr.io/google_containers/kubernetes-dashboard-amd64:v1.10.1 jiashiwen/kubernetes-dashboard-amd64:v1.10.1
  docker push jiashiwen/kubernetes-dashboard-amd64:v1.10.1
  
  docker pull gcr.io/google_containers/kube-apiserver:v1.14.1
  docker tag gcr.io/google_containers/kube-apiserver:v1.14.1 jiashiwen/kube-apiserver:v1.14.1
  docker push jiashiwen/kube-apiserver:v1.14.1
  
  docker pull gcr.io/google_containers/kube-controller-manager:v1.14.1
  docker tag gcr.io/google_containers/kube-controller-manager:v1.14.1 jiashiwen/kube-controller-manager:v1.14.1
  docker push jiashiwen/kube-controller-manager:v1.14.1
  
  docker pull gcr.io/google_containers/kube-scheduler:v1.14.1
  docker tag gcr.io/google_containers/kube-scheduler:v1.14.1 jiashiwen/kube-scheduler:v1.14.1
  docker push jiashiwen/kube-scheduler:v1.14.1
  
  docker pull gcr.io/google_containers/kube-proxy:v1.14.1
  docker tag gcr.io/google_containers/kube-proxy:v1.14.1 jiashiwen/kube-proxy:v1.14.1
  docker push jiashiwen/kube-proxy:v1.14.1
  
  docker pull gcr.io/google_containers/pause:3.1
  docker tag gcr.io/google_containers/pause:3.1 jiashiwen/pause:3.1
  docker push jiashiwen/pause:3.1
  
  docker pull gcr.io/google_containers/coredns:1.3.1
  docker tag gcr.io/google_containers/coredns:1.3.1 jiashiwen/coredns:1.3.1
  docker push  jiashiwen/coredns:1.3.1
  

• 用于下载上传镜像的脚本

  #!/bin/bash
  
  privaterepo=jiashiwen
  
  k8sgcrimages=(
  cluster-proportional-autoscaler-amd64:1.4.0
  k8s-dns-node-cache:1.15.1
  )
  
  gcrimages=(
  pause-amd64:3.1
  kubernetes-dashboard-amd64:v1.10.1
  kube-apiserver:v1.14.1
  kube-controller-manager:v1.14.1
  kube-scheduler:v1.14.1
  kube-proxy:v1.14.1
  pause:3.1
  coredns:1.3.1
  )
  
  
  for k8sgcrimageName in ${k8sgcrimages[@]} ; do
  echo $k8sgcrimageName
  docker pull k8s.gcr.io/$k8sgcrimageName
  docker tag k8s.gcr.io/$k8sgcrimageName $privaterepo/$k8sgcrimageName
  docker push $privaterepo/$k8sgcrimageName
  done
  
  
  for gcrimageName in ${gcrimages[@]} ; do
  echo $gcrimageName
  docker pull gcr.io/google_containers/$gcrimageName
  docker tag gcr.io/google_containers/$gcrimageName $privaterepo/$gcrimageName
  docker push $privaterepo/$gcrimageName
  done
  

• 修改文件inventory/mycluster/group_vars/k8s-cluster/k8s-cluster.yml，修改k8s镜像仓库

  # kube_image_repo: "gcr.io/google-containers"
  kube_image_repo: "jiashiwen"
  

• 修改roles/download/defaults/main.yml

  #dnsautoscaler_image_repo: "k8s.gcr.io/cluster-proportional-autoscaler-{{   image_arch }}"
  dnsautoscaler_image_repo: "jiashiwen/cluster-proportional-autoscaler-{{   image_arch }}"
  
  #kube_image_repo: "gcr.io/google-containers"
  kube_image_repo: "jiashiwen"
  
  #pod_infra_image_repo: "gcr.io/google_containers/pause-{{ image_arch }}"
  pod_infra_image_repo: "jiashiwen/pause-{{ image_arch }}"
  
  #dashboard_image_repo: "gcr.io/google_containers/kubernetes-dashboard-{{   image_arch }}"
  dashboard_image_repo: "jiashiwen/kubernetes-dashboard-{{ image_arch }}"
  
  #nodelocaldns_image_repo: "k8s.gcr.io/k8s-dns-node-cache"
  nodelocaldns_image_repo: "jiashiwen/k8s-dns-node-cache"
  
  #kubeadm_download_url: "https://storage.googleapis.com/kubernetes-release/  release/{{ kubeadm_version }}/bin/linux/{{ image_arch }}/kubeadm"
  kubeadm_download_url: "http://10.0.0.60:8888/kubeadm"
  
  #hyperkube_download_url: "https://storage.googleapis.com/  kubernetes-release/release/{{ kube_version }}/bin/linux/{{ image_arch }}/  hyperkube"
  hyperkube_download_url: "http://10.0.0.60:8888/hyperkube"
  

执行安装

• 安装命令

  ansible-playbook -i inventory/mycluster/inventory.ini cluster.yml
  

• 重置命令

  ansible-playbook -i inventory/mycluster/inventory.ini reset.yml
  

验证k8s集群

• 安装kubectl

  • ubuntu

    sudo snap install kubectl --classic
    

  • centos

    • 本地浏览器打开 https://storage.googleapis.com/kubernetes-release/release/stable.txt得到最新版本为v1.14.1

    • 用上一步得到的最新版本号v1.7.1替换下载地址中的$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt )得到真正的下载地址https:// storage.googleapis.com/kubernetes-release/release/v1.14.1/bin/linux/amd64/kubectl

    • 上传下载好的kubectl

      scp /tmp/kubectl root@xxx:/root
      

    • 修改属性

      chmod +x ./kubectl
      mv ./kubectl /usr/local/bin/kubectl
      

• 将master节点上的~/.kube/config 文件复制到你需要访问集群的客户端上即可

  scp 10.0.0.40:/root/.kube/config ~/.kube/config
  

• 执行命令验证集群

  kubectl get nodes
  kubectl cluster-info
  

tidb-operaor部署

安装helm

https://blog.csdn.net/bbwangj/article/details/81087911

• 安装helm

  curl https://raw.githubusercontent.com/helm/helm/master/scripts/get > get_helm.sh
  chmod 700 get_helm.sh
  ./get_helm.sh
  

• 查看helm版本

  helm version
  

• 初始化

  helm init --upgrade -i registry.cn-hangzhou.aliyuncs.com/google_containers/tiller:v2.13.1 --stable-repo-url https://kubernetes.oss-cn-hangzhou.aliyuncs.com/charts
  

为k8s提供 local volumes

• 参考文档 https://github.com/kubernetes-sigs/sig-storage-local-static-provisioner/blob/master/docs/operations.md tidb-operator启动会为pd和tikv绑定pv，需要在discovery directory下创建多个目录

• 格式化并挂载磁盘

  mkfs.ext4 /dev/vdb
  DISK_UUID=$(blkid -s UUID -o value /dev/vdb) 
  mkdir /mnt/$DISK_UUID
  mount -t ext4 /dev/vdb /mnt/$DISK_UUID
  

• /etc/fstab持久化mount

  echo UUID=`sudo blkid -s UUID -o value /dev/vdb` /mnt/$DISK_UUID ext4 defaults 0 2 | sudo tee -a /etc/fstab
  

• 创建多个目录并mount到discovery directory

  for i in $(seq 1 10); do
  sudo mkdir -p /mnt/${DISK_UUID}/vol${i} /mnt/disks/${DISK_UUID}_vol${i}
  sudo mount --bind /mnt/${DISK_UUID}/vol${i} /mnt/disks/${DISK_UUID}_vol${i}
  done
  

• /etc/fstab持久化mount

  for i in $(seq 1 10); do
  echo /mnt/${DISK_UUID}/vol${i} /mnt/disks/${DISK_UUID}_vol${i} none bind 0 0 | sudo tee -a /etc/fstab
  done
  

• 为tidb-operator创建local-volume-provisioner

$ kubectl apply -f https://raw.githubusercontent.com/pingcap/tidb-operator/master/manifests/local-dind/local-volume-provisioner.yaml
$ kubectl get po -n kube-system -l app=local-volume-provisioner
$ kubectl get pv --all-namespaces | grep local-storage 

Install TiDB Operator

• 项目中使用了gcr.io/google-containers/hyperkube，国内访问不了，简单的办法是把镜像重新push到dockerhub然后修改charts/tidb-operator/values.yaml

  scheduler:
    # With rbac.create=false, the user is responsible for creating this   account
    # With rbac.create=true, this service account will be created
    # Also see rbac.create and clusterScoped
    serviceAccount: tidb-scheduler
    logLevel: 2
    replicas: 1
    schedulerName: tidb-scheduler
    resources:
      limits:
        cpu: 250m
        memory: 150Mi
      requests:
        cpu: 80m
        memory: 50Mi
    # kubeSchedulerImageName: gcr.io/google-containers/hyperkube
    kubeSchedulerImageName: yourrepo/hyperkube
    # This will default to matching your kubernetes version
    # kubeSchedulerImageTag: latest
  

• TiDB Operator使用CRD扩展Kubernetes，因此要使用TiDB Operator，首先应该创建TidbCluster自定义资源类型。

  kubectl apply -f https://raw.githubusercontent.com/pingcap/tidb-operator/master/manifests/crd.yaml
  kubectl get crd tidbclusters.pingcap.com
  

• 安装tidb-operator

  $ git clone https://github.com/pingcap/tidb-operator.git
  $ cd tidb-operator
  $ helm install charts/tidb-operator --name=tidb-operator   --namespace=tidb-admin
  $ kubectl get pods --namespace tidb-admin -l app.kubernetes.io/  instance=tidb-operator
  

部署tidb

helm install charts/tidb-cluster --name=demo --namespace=tidb
watch kubectl get pods --namespace tidb -l app.kubernetes.io/instance=demo -o wide

验证

安装mysql客户端

• 参考文档 https://dev.mysql.com/doc/refman/8.0/en/linux-installation.html

• centos安装

  wget https://dev.mysql.com/get/mysql80-community-release-el7-3.noarch.rpm
  yum localinstall mysql80-community-release-el7-3.noarch.rpm -y
  yum repolist all | grep mysql
  yum-config-manager --disable mysql80-community
  yum-config-manager --enable mysql57-community
  yum install mysql-community-client
  

• ubuntu安装

  wget https://dev.mysql.com/get/mysql-apt-config_0.8.13-1_all.deb
  dpkg -i mysql-apt-config_0.8.13-1_all.deb
  apt update
  
  # 选择mysql版本
  dpkg-reconfigure mysql-apt-config
  apt install mysql-client -y
  

映射tidb端口

• 查看tidb serveice

  kubectl get svc --all-namespaces
  

• 映射tidb端口

  # 仅本地访问
  kubectl port-forward svc/demo-tidb 4000:4000 --namespace=tidb
  
  # 其他主机访问
  kubectl port-forward --address 0.0.0.0 svc/demo-tidb 4000:4000 --namespace=tidb
  

• 首次登录mysql

  mysql -h 127.0.0.1 -P 4000 -u root -D test
  

• 修改tidb密码

  SET PASSWORD FOR 'root'@'%' = 'wD3cLpyO5M'; FLUSH PRIVILEGES;
  

趟坑小记

• k8s国内安装 k8s镜像多在gcr.io国内访问不到，基本做法是把镜像导入dockerhub或者私有镜像，这一点在k8s部署章节有详细过程就不累述了。
• tidb-operator 本地存储配置 operator在启动集群时pd和tikv需要绑定本地存储如果挂载点不足会导致pod启动过程中找不到可已bond的pv始终处于pending或createing状态，详细配请参阅 https://github.com/kubernetes-sigs/sig-storage-local-static-provisioner/blob/master/docs/operations.md中“Sharing a disk filesystem by multiple filesystem PVs”一节，同一块磁盘绑定多个挂载目录，为operator提供足够的bond
• mysql客户端版本问题 目前tidb只支持mysql5.7版本客户端8.0会报ERROR 1105 (HY000): Unknown charset id 255