springboot使用zookeeper(curator)实现注册发现与负载均衡

最简单的实现服务高可用的方法就是集群化,也就是分布式部署,但是分布式部署会带来一些问题。比如:

1、各个实例之间的协同(锁)

2、负载均衡

3、热删除

这里通过一个简单的实例来说明如何解决注册发现和负载均衡。

1、先解决依赖,这里只给出zk相关的依赖,pom.xml如下

 <dependency>
    <groupId>org.apache.zookeeper</groupId>
    <artifactId>zookeeper</artifactId>
    <version>3.4.8</version>
</dependency>
<dependency>
    <groupId>org.apache.curator</groupId>
    <artifactId>curator-recipes</artifactId>
    <version>2.9.1</version>
</dependency>
<dependency>
    <groupId>org.apache.curator</groupId>
    <artifactId>curator-client</artifactId>
    <version>2.9.1</version>
  </dependency>

2、ZkClient

这里使用的是curator,curator是对zookeeper的简单封装,提供了一些集成的方法,或者是提供了更优雅的api,举例来说

zk的create(path, mode, acl, data)方法 == curator create().withMode(mode).forPath(path)调用链

import org.apache.curator.RetryPolicy;
import org.apache.curator.framework.CuratorFramework;
import org.apache.curator.framework.CuratorFrameworkFactory;
import org.apache.curator.retry.ExponentialBackoffRetry;
import org.apache.zookeeper.CreateMode;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import java.net.InetAddress;
import java.util.ArrayList;
import java.util.List;

public class ZkClient {
     
    private final Logger logger = LoggerFactory.getLogger(this.getClass());

    private CuratorFramework client;
    private String zookeeperServer;
    private int sessionTimeoutMs;
    private int connectionTimeoutMs;
    private int baseSleepTimeMs;
    private int maxRetries;

    public void setZookeeperServer(String zookeeperServer) {
     
        this.zookeeperServer = zookeeperServer;
    }
    public String getZookeeperServer() {
     
        return zookeeperServer;
    }
    public void setSessionTimeoutMs(int sessionTimeoutMs) {
     
        this.sessionTimeoutMs = sessionTimeoutMs;
    }
    public int getSessionTimeoutMs() {
     
        return sessionTimeoutMs;
    }
    public void setConnectionTimeoutMs(int connectionTimeoutMs) {
     
        this.connectionTimeoutMs = connectionTimeoutMs;
    }
    public int getConnectionTimeoutMs() {
     
        return connectionTimeoutMs;
    }
    public void setBaseSleepTimeMs(int baseSleepTimeMs) {
     
        this.baseSleepTimeMs = baseSleepTimeMs;
    }
    public int getBaseSleepTimeMs() {
     
        return baseSleepTimeMs;
    }
    public void setMaxRetries(int maxRetries) {
     
        this.maxRetries = maxRetries;
    }
    public int getMaxRetries() {
     
        return maxRetries;
    }

    public void init() {
     
        RetryPolicy retryPolicy = new ExponentialBackoffRetry(baseSleepTimeMs, maxRetries);
        client = CuratorFrameworkFactory.builder().connectString(zookeeperServer).retryPolicy(retryPolicy)
                .sessionTimeoutMs(sessionTimeoutMs).connectionTimeoutMs(connectionTimeoutMs).build();
        client.start();
    }

    public void stop() {
     
        client.close();
    }

    public CuratorFramework getClient() {
     
        return client;
    }

    public void register() {
     
        try {
     
            String rootPath = "/" + "services";
            String hostAddress = InetAddress.getLocalHost().getHostAddress();
            String serviceInstance = "prometheus" + "-" +  hostAddress + "-";
            client.create().creatingParentsIfNeeded().withMode(CreateMode.EPHEMERAL_SEQUENTIAL).forPath(rootPath + "/" + serviceInstance);
        } catch (Exception e) {
     
            logger.error("注册出错", e);
        }
    }

    public List<String> getChildren(String path) {
     
        List<String> childrenList = new ArrayList<>();
        try {
     
            childrenList = client.getChildren().forPath(path);
        } catch (Exception e) {
     
            logger.error("获取子节点出错", e);
        }
        return childrenList;
    }

    public int getChildrenCount(String path) {
     
        return getChildren(path).size();
    }

    public List<String> getInstances() {
     
        return getChildren("/services");
    }

    public int getInstancesCount() {
     
        return getInstances().size();
    }
}

2、configuration如下

import com.dqa.prometheus.client.zookeeper.ZkClient;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

@Configuration
public class ZkConfiguration {
     
    @Value("${zookeeper.server}")
    private String zookeeperServer;
    @Value(("${zookeeper.sessionTimeoutMs}"))
    private int sessionTimeoutMs;
    @Value("${zookeeper.connectionTimeoutMs}")
    private int connectionTimeoutMs;
    @Value("${zookeeper.maxRetries}")
    private int maxRetries;
    @Value("${zookeeper.baseSleepTimeMs}")
    private int baseSleepTimeMs;

    @Bean(initMethod = "init", destroyMethod = "stop")
    public ZkClient zkClient() {
     
        ZkClient zkClient = new ZkClient();
        zkClient.setZookeeperServer(zookeeperServer);
        zkClient.setSessionTimeoutMs(sessionTimeoutMs);
        zkClient.setConnectionTimeoutMs(connectionTimeoutMs);
        zkClient.setMaxRetries(maxRetries);
        zkClient.setBaseSleepTimeMs(baseSleepTimeMs);
        return zkClient;
    }

}

配置文件如下

#============== zookeeper ===================
zookeeper.server=10.93.21.21:2181,10.93.18.34:2181,10.93.18.35:2181
zookeeper.sessionTimeoutMs=6000
zookeeper.connectionTimeoutMs=6000
zookeeper.maxRetries=3
zookeeper.baseSleepTimeMs=1000
 

3、注册发现

是通过上面封装的ZkClient中的register方法实现的,调用如下。

import com.dqa.prometheus.client.zookeeper.ZkClient;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
import org.springframework.boot.orm.jpa.EntityScan;
import org.springframework.context.ApplicationContext;
import org.springframework.scheduling.annotation.EnableAsync;
import org.springframework.scheduling.annotation.EnableScheduling;

@SpringBootApplication
@EnableAsync
@EnableScheduling
@EntityScan(basePackages="com.xiaoju.dqa.prometheus.model")
public class Application {
     
    public static void main(String[] args) {
     
        ApplicationContext context = SpringApplication.run(Application.class, args);
        ZkClient zkClient = context.getBean(ZkClient.class);
        zkClient.register();
    }
}

1、zk中的注册路径

/services/prometheus-10.93.21.21-00000000001

2、CreateMode有四种,选择EPHEMERAL_SEQUENTIAL的原因是,服务关闭的时候session超时,zk节点会自动删除,同时自增id可以实现锁和负载均衡,下面再说

1、PERSISTENT

持久化目录节点,存储的数据不会丢失。

2、PERSISTENT_SEQUENTIAL

顺序自动编号的持久化目录节点,存储的数据不会丢失,并且根据当前已近存在的节点数自动加 1,然后返回给客户端已经成功创建的目录节点名。

3、EPHEMERAL

临时目录节点,一旦创建这个节点的客户端与服务器端口也就是session 超时,这种节点会被自动删除。

4、EPHEMERAL_SEQUENTIAL

临时自动编号节点,一旦创建这个节点的客户端与服务器端口也就是session 超时,这种节点会被自动删除,并且根据当前已近存在的节点数自动加 1,然后返回给客户端已经成功创建的目录节点名。
复制代码
4、负载均衡

   /*
*   我是第几个实例, 做负载均衡
* */
List<String> instanceList = zkClient.getInstances();
Collections.sort(instanceList);
String hostAddress = NetFunction.getAddressHost();
int instanceNo = 0;
if (hostAddress !=  null) {
     
    for (int i=0; i<instanceList.size(); i++) {
     
        if (instanceList.get(i).split("-")[1].equals(hostAddress)) {
     
            instanceNo = i;
        }
    }
} else {
     
    logger.info("获取本地IP失败");
}
logger.info("[分发] 实例总数={}, 我是第{}个实例", instanceCount, instanceNo);
List<CheckTask> waitingTasks = checkTaskDao.getTasks(taskType, TaskStatus.WAITING.getValue());
Iterator<CheckTask> waitingIterator = waitingTasks.iterator();
while (waitingIterator.hasNext()) {
     
    if (waitingIterator.next().getTaskId().hashCode() % instanceCount != instanceNo) {
     
        waitingIterator.remove();
    }
}

说明:

1、例如有3个实例(zkClient.getInstances()),那么通过IP我们把3个实例按照自增id排序分别标号为0,1,2

2、对第一个实例也就是instanceNo=0,只执行taskId.hashCode() % 3 == 0的任务,其他两个实例类似

3、当有一个实例挂掉,2个实例,instanceNo=0只执行taskId.hashCode() % 2 == 0的任务,实现热删除

你可能感兴趣的:(代码,注解,配置,java,zookeeper)