fastdfs分布式文件系统之TrackerServer连接池实现

公司使用fastdfs文件系统来存储文件和图片，为了避免每个系统都直接通过客户端直接访问fastdfs文件系统，所以我们做了一个

dubbo服务。要实现图片上传、下载的服务或工程直接调用dubbo接口就可以了。为了提高性能对fastdfs文件系统TrackerServer写了个

连接池，同时写了空闲连接是否可用的监测类。

下面对连接池的具体实现作介绍。

连接池类

实现配置文件的读取、初始化各项参数、创建默认的TrackerServer数、空闲连接获取、释放繁忙连接及删除不可用的连接；

package com.xxxx.fastdfs.pool;import java.io.IOException;import java.util.UUID;import java.util.concurrent.LinkedBlockingQueue;import java.util.concurrent.TimeUnit;import org.csource.fastdfs.ClientGlobal;import org.csource.fastdfs.TrackerClient;import org.csource.fastdfs.TrackerServer;import org.slf4j.Logger;import org.slf4j.LoggerFactory;import com.xxxx.common.AppException;import com.xxxx.fastdfs.exception.ERRORS;/** *  * @ClassName: ConnectionPool * @Description: fastdfs连接池 * @author mr_smile2014 [e-mail] * @date 2015年10月26日 下午3:36:02 *  */public class ConnectionPool {private static final Logger LOGGER = LoggerFactory.getLogger(ConnectionPool.class);/** 空闲的连接池 */private LinkedBlockingQueue<TrackerServer> idleConnectionPool = null;/** 连接池默认最小连接数 */private long minPoolSize = 10;/** 连接池默认最大连接数 */private long maxPoolSize = 30;/** 当前创建的连接数 */private volatile long nowPoolSize = 0;/** 默认等待时间（单位：秒） */private long waitTimes = 200;/** fastdfs客户端创建连接默认1次 */private static final int COUNT = 1;/** * 默认构造方法 */public ConnectionPool(long minPoolSize, long maxPoolSize, long waitTimes) {String logId = UUID.randomUUID().toString();LOGGER.info("[线程池构造方法(ConnectionPool)][" + logId+ "][默认参数：minPoolSize=" + minPoolSize + ",maxPoolSize="+ maxPoolSize + ",waitTimes=" + waitTimes + "]");this.minPoolSize = minPoolSize;this.maxPoolSize = maxPoolSize;this.waitTimes = waitTimes;/** 初始化连接池 */poolInit(logId);/** 注册心跳 */HeartBeat beat = new HeartBeat(this);beat.beat();}/** *  * @Description: 连接池初始化 (在加载当前ConnectionPool时执行) 1).加载配置文件 2).空闲连接池初始化； *               3).创建最小连接数的连接，并放入到空闲连接池； *  */private void poolInit(String logId) {try {/** 加载配置文件 */initClientGlobal();/** 初始化空闲连接池 */idleConnectionPool = new LinkedBlockingQueue<TrackerServer>();/** 往线程池中添加默认大小的线程 */for (int i = 0; i < minPoolSize; i++) {createTrackerServer(logId, COUNT);}} catch (Exception e) {LOGGER.error("[FASTDFS初始化(init)--异常][" + logId + "][异常：{}]", e);}}/** *  * @Description: 创建TrackerServer,并放入空闲连接池 *  */public void createTrackerServer(String logId, int flag) {LOGGER.info("[创建TrackerServer(createTrackerServer)][" + logId + "]");TrackerServer trackerServer = null;try {TrackerClient trackerClient = new TrackerClient();trackerServer = trackerClient.getConnection();while (trackerServer == null && flag < 5) {LOGGER.info("[创建TrackerServer(createTrackerServer)][" + logId+ "][第" + flag + "次重建]");flag++;initClientGlobal();trackerServer = trackerClient.getConnection();}org.csource.fastdfs.ProtoCommon.activeTest(trackerServer.getSocket());idleConnectionPool.add(trackerServer);/** 同一时间只允许一个线程对nowPoolSize操作 **/synchronized (this) {nowPoolSize++;}} catch (Exception e) {LOGGER.error("[创建TrackerServer(createTrackerServer)][" + logId+ "][异常：{}]", e);} finally {if (trackerServer != null) {try {trackerServer.close();} catch (Exception e) {LOGGER.error("[创建TrackerServer(createTrackerServer)--关闭trackerServer异常]["+ logId + "][异常：{}]", e);}}}}/** *  * @Description: 获取空闲连接 1).在空闲池（idleConnectionPool)中弹出一个连接； *               2).把该连接放入忙碌池（busyConnectionPool）中; 3).返回 connection *               4).如果没有idle connection, 等待 wait_time秒, and check again *  * @throws AppException *  */public TrackerServer checkout(String logId) throws AppException {LOGGER.info("[获取空闲连接(checkout)][" + logId + "]");TrackerServer trackerServer = idleConnectionPool.poll();if (trackerServer == null) {if (nowPoolSize < maxPoolSize) {createTrackerServer(logId, COUNT);try {trackerServer = idleConnectionPool.poll(waitTimes,TimeUnit.SECONDS);} catch (Exception e) {LOGGER.error("[获取空闲连接(checkout)-error][" + logId+ "][error:获取连接超时:{}]",e);throw ERRORS.WAIT_IDLECONNECTION_TIMEOUT.ERROR();}}if (trackerServer == null) {LOGGER.error("[获取空闲连接(checkout)-error][" + logId+ "][error:获取连接超时（" + waitTimes + "s）]");throw ERRORS.WAIT_IDLECONNECTION_TIMEOUT.ERROR();}}LOGGER.info("[获取空闲连接(checkout)][" + logId + "][获取空闲连接成功]");return trackerServer;}/** *  * @Description: 释放繁忙连接 1.如果空闲池的连接小于最小连接值，就把当前连接放入idleConnectionPool； *               2.如果空闲池的连接等于或大于最小连接值，就把当前释放连接丢弃； *  * @param client1 *            需释放的连接对象 *  */public void checkin(TrackerServer trackerServer, String logId) {LOGGER.info("[释放当前连接(checkin)][" + logId + "][prams:" + trackerServer+ "] ");if (trackerServer != null) {if (idleConnectionPool.size() < minPoolSize) {idleConnectionPool.add(trackerServer);} else {synchronized (this) {if (nowPoolSize != 0) {nowPoolSize--;}}}}}/** *  * @Description: 删除不可用的连接，并把当前连接数减一（调用过程中trackerServer报异常，调用一般在finally中） * @param trackerServer *  */public void drop(TrackerServer trackerServer, String logId) {LOGGER.info("[删除不可用连接方法(drop)][" + logId + "][parms:" + trackerServer+ "] ");if (trackerServer != null) {try {synchronized (this) {if (nowPoolSize != 0) {nowPoolSize--;}}trackerServer.close();} catch (IOException e) {LOGGER.info("[删除不可用连接方法(drop)--关闭trackerServer异常][" + logId+ "][异常：{}]", e);}}}private void initClientGlobal() throws Exception {ClientGlobal.init("fastdfs_client.conf");}public LinkedBlockingQueue<TrackerServer> getIdleConnectionPool() {return idleConnectionPool;}public long getMinPoolSize() {return minPoolSize;}public void setMinPoolSize(long minPoolSize) {if (minPoolSize != 0) {this.minPoolSize = minPoolSize;}}public long getMaxPoolSize() {return maxPoolSize;}public void setMaxPoolSize(long maxPoolSize) {if (maxPoolSize != 0) {this.maxPoolSize = maxPoolSize;}}public long getWaitTimes() {return waitTimes;}public void setWaitTimes(int waitTimes) {if (waitTimes != 0) {this.waitTimes = waitTimes;}}}

连接池定时检测器

用于定时检测连接池空闲队列中的连接是否可用，如果不可用将从连接池中移除。

package com.xxxx.fastdfs.pool;import java.util.Timer;import java.util.TimerTask;import java.util.UUID;import java.util.concurrent.LinkedBlockingQueue;import java.util.concurrent.TimeUnit;import org.csource.fastdfs.TrackerServer;import org.slf4j.Logger;import org.slf4j.LoggerFactory;/** *  * @ClassName: HeartBeat * @Description: 连接池定时器设置 * @author mr_smile2014 [e-mail] * @date 2015年1月14日 下午2:21:29 *  */public class HeartBeat {private static final Logger LOGGER = LoggerFactory.getLogger(HeartBeat.class);/** fastdfs连接池 */private ConnectionPool pool = null;/** 小时毫秒数 */public static int ahour = 1000 * 60 * 60 * 1;/** 等待时间 */public static int waitTimes = 200;public HeartBeat(ConnectionPool pool) {this.pool = pool;}/** *  * @Description: 定时执行任务，检测当前的空闲连接是否可用，如果不可用将从连接池中移除 *  */public void beat() {LOGGER.info("[心跳任务方法（beat）]");TimerTask task = new TimerTask() {@Overridepublic void run() {String logId = UUID.randomUUID().toString();LOGGER.info("[心跳任务方法（beat）]["+ logId+ "][Description:对idleConnectionPool中的TrackerServer进行监测]");LinkedBlockingQueue<TrackerServer> idleConnectionPool = pool.getIdleConnectionPool();TrackerServer ts = null;for (int i = 0; i < idleConnectionPool.size(); i++) {try {ts = idleConnectionPool.poll(waitTimes,TimeUnit.SECONDS);if (ts != null) {org.csource.fastdfs.ProtoCommon.activeTest(ts.getSocket());idleConnectionPool.add(ts);} else {/** 代表已经没有空闲长连接 */break;}} catch (Exception e) {/** 发生异常,要删除，进行重建 */LOGGER.error("[心跳任务方法（beat）][" + logId+ "][异常：当前连接已不可用将进行重新获取连接]");pool.drop(ts, logId);}}}};Timer timer = new Timer();timer.schedule(task, ahour, ahour);}}

连接池调用

/** 获取fastdfs服务器连接 */trackerServer = connectionPool.checkout(logId);StorageServer storageServer = null;StorageClient1 client1 = new StorageClient1(trackerServer,storageServer);/** 以文件字节的方式上传 */........................../** 上传完毕及时释放连接 */connectionPool.checkin(trackerServer, logId);

连接池的使用最好是再写一个工具类，在工具类当中进行连接池的初始化及上传、下载、删除功能的实现。把工具类配置到spring配置文件中，

生成对应的bean,其它用到的地方直接调用该bean的对应方法就可以了。工具类的实现将在下一篇文章中分享。