Kubernetes Operator开发与StatefulSet有状态应用管理

背景与挑战

在Kubernetes生态中，有状态应用（如数据库、消息队列、缓存系统）的管理一直是最具挑战性的领域之一。与传统无状态应用相比，有状态应用需要确保：

• 数据持久性：确保数据不因Pod重启或迁移而丢失
• 稳定标识：维护固定的网络标识和存储映射关系
• 有序操作：部署、扩展、升级等操作需要严格的顺序控制
• 状态一致性：保证集群状态的一致性和数据完整性

以MySQL数据库集群为例，传统Deployment无法满足以下需求：
- 主从节点的有序启动和配置
- 自动故障转移和数据同步
- 备份恢复的自动化管理
- 配置变更的滚动更新

实现方案

基于Operator模式我们开发了MySQL集群管理Operator，实现有状态应用的自动化管理

• CRD（CustomResourceDefinition）​：定义MySQLCluster资源类型，封装集群规格、备份策略等配置
• 控制器逻辑​​：持续调和实际状态与期望状态，处理节点故障转移、数据备份等复杂操作
• 初始化与配置管理​​：自动处理MySQL系统初始化、主从复制配置和参数优化

apiVersion: mysql.oracle.com/v2
kind: InnoDBCluster
metadata:
  name: mycluster
spec:
  instances: 3
  secretName: my-secret
  tlsUseSelfSigned: true
  router:
    instances: 1
  podSpec:
    resources:
      requests:
        memory: "2Gi"
        cpu: "1000m"
      limits:
        memory: "4Gi" 
        cpu: "2000m"
    # 关键：Pod反亲和性确保节点分布
    affinity:
      podAntiAffinity:
        requiredDuringSchedulingIgnoredDuringExecution:
   

Before You Begin

SSH Configuration

vi /etc/hosts
x.x.x.x linode01
x.x.x.x linode02
x.x.x.x linode03

hostnamectl set-hostname linode01

ssh-keygen -t rsa
ssh-copy-id -i ~/.ssh/id_rsa.pub linode01
ssh-copy-id -i ~/.ssh/id_rsa.pub linode02
ssh-copy-id -i ~/.ssh/id_rsa.pub linode03

Disable Firewall

systemctl stop firewalld.service
systemctl disable firewalld.service

Dependency

yum install psmisc -y
yum install libxslt-devel -y
yum install chkconfig bind-utils psmisc libxslt zlib sqlite cyrus-sasl-plain cyrus-sasl-gssapi fuse portmap fuse-libs redhat-lsb -y
yum install python-psycopg2 -y 
yum install snappy snappy-devel  -y

#NFS
yum install rpcbind -y
service rpcbind start

Install the Oracle JDK

cd /opt/
sudo wget --no-cookies --no-check-certificate --header "Cookie: gpw_e24=http%3A%2F%2Fwww.oracle.com%2F; oraclelicense=accept-securebackup-cookie" "http://download.oracle.com/otn/java/jdk/7u80-b15/jdk-7u80-linux-x64

Before You Begin

SSH Configuration

vi /etc/hosts
x.x.x.x linode01
x.x.x.x linode02
x.x.x.x linode03

hostnamectl set-hostname linode01

ssh-keygen -t rsa
ssh-copy-id -i ~/.ssh/id_rsa.pub linode01
ssh-copy-id -i ~/.ssh/id_rsa.pub linode02
ssh-copy-id -i ~/.ssh/id_rsa.pub linode03

Disable Firewall

systemctl stop firewalld.service
systemctl disable firewalld.service

Dependency

yum install psmisc -y
yum install libxslt-devel -y
yum install chkconfig bind-utils psmisc libxslt zlib sqlite cyrus-sasl-plain cyrus-sasl-gssapi fuse portmap fuse-libs redhat-lsb -y
yum install python-psycopg2 -y 
yum install snappy snappy-devel  -y

#NFS
yum install rpcbind -y
service rpcbind start

Install and Configure External Databases

sudo yum install postgresql-server postgresql -y
sudo su - postgres
initdb -D /var/lib/pgsql/data

#remote access
vi /var/lib/pgsql/data/postgresql.conf
listen_addresses ='*'

vi /var/lib/pgsql/data/pg_h