Druid核心源码解析--DruidDataSource

Posted 2022-10-24 水田如雅

配置读取

druid连接池支持的所有连接参数可在类com.alibaba.druid.pool.DruidDataSourceFactory中查看。

配置读取代码：

 public void configFromPropety(Properties properties) 
        //这方法太长，自己看源码去吧，就是读读属性。。。。
    

整体代码比较简单，就是把配置内容，读取到dataSource。

连接池初始化

首先是简单的判断，加锁：

if (inited) 
            //已经被初始化好了，直接return
            return;
        

        // bug fixed for dead lock, for issue #2980
        DruidDriver.getInstance();
        /**控制创建移除连接的锁，并且通过条件去控制一个连接的生成消费**/
        // public DruidAbstractDataSource(boolean lockFair)
        //        lock = new ReentrantLock(lockFair);
        //
        //        notEmpty = lock.newCondition();
        //        empty = lock.newCondition();
        //    
        final ReentrantLock lock = this.lock;
        try 
            lock.lockInterruptibly();
         catch (InterruptedException e) 
            throw new SQLException("interrupt", e);
        

之后会更新一些JMX的监控指标：

//一些jmx监控指标
                this.connectionIdSeedUpdater.addAndGet(this, delta);
                this.statementIdSeedUpdater.addAndGet(this, delta);
                this.resultSetIdSeedUpdater.addAndGet(this, delta);
                this.transactionIdSeedUpdater.addAndGet(this, delta);

druid的监控指标都是通过jmx实现的。

解析连接串：

 if (this.jdbcUrl != null) 
                //解析连接串
                this.jdbcUrl = this.jdbcUrl.trim();
                initFromWrapDriverUrl();
            

initFromWrapDriverUrl方法，除了从jdbc url中解析出连接和驱动信息，后面还把filters的名字，解析成了对应的filter类。

  private void initFromWrapDriverUrl() throws SQLException 
        if (!jdbcUrl.startsWith(DruidDriver.DEFAULT_PREFIX)) 
            return;
        

        DataSourceProxyConfig config = DruidDriver.parseConfig(jdbcUrl, null);
        this.driverClass = config.getRawDriverClassName();

        LOG.error("error url : '" + jdbcUrl + "', it should be : '" + config.getRawUrl() + "'");

        this.jdbcUrl = config.getRawUrl();
        if (this.name == null) 
            this.name = config.getName();
        

        for (Filter filter : config.getFilters()) 
            addFilter(filter);
        
    

之后在init方法里面，会进行filters的初始化：

 //初始化filter 属性
            for (Filter filter : filters) 
                filter.init(this);
            

之后解析数据库类型：

 if (this.dbTypeName == null || this.dbTypeName.length() == 0) 
                this.dbTypeName = JdbcUtils.getDbType(jdbcUrl, null);
            

注意枚举值： com.alibaba.druid.DbType，这个里面包含了目前durid连接池支持的所有数据源 类型，另外，druid还额外提供了一些驱动类，例如：

 elastic_search  (1 << 25), // com.alibaba.xdriver.elastic.jdbc.ElasticDriver

clickhouse还提供了负载均衡的驱动类：

com.alibaba.druid.support.clickhouse.BalancedClickhouseDriver 。

在回到init方法，之后是一堆参数解析，不再说，跳过了。
之后是通过SPI加载自定义的filter:

  private void initFromSPIServiceLoader() 
        if (loadSpifilterSkip) 
            return;
        

        if (autoFilters == null) 
            List<Filter> filters = new ArrayList<Filter>();
            ServiceLoader<Filter> autoFilterLoader = ServiceLoader.load(Filter.class);

            for (Filter filter : autoFilterLoader) 
                AutoLoad autoLoad = filter.getClass().getAnnotation(AutoLoad.class);
                if (autoLoad != null && autoLoad.value()) 
                    filters.add(filter);
                
            
            autoFilters = filters;
        

        for (Filter filter : autoFilters) 
            if (LOG.isInfoEnabled()) 
                LOG.info("load filter from spi :" + filter.getClass().getName());
            
            addFilter(filter);
        
    

注意自定义的filter,要使用com.alibaba.druid.filter.AutoLoad。

解析驱动：

  protected void resolveDriver() throws SQLException 
        if (this.driver == null) 
            if (this.driverClass == null || this.driverClass.isEmpty()) 
                this.driverClass = JdbcUtils.getDriverClassName(this.jdbcUrl);
            

            if (MockDriver.class.getName().equals(driverClass)) 
                driver = MockDriver.instance;
             else if ("com.alibaba.druid.support.clickhouse.BalancedClickhouseDriver".equals(driverClass)) 
                Properties info = new Properties();
                info.put("user", username);
                info.put("password", password);
                info.putAll(connectProperties);
                driver = new BalancedClickhouseDriver(jdbcUrl, info);
             else 
                if (jdbcUrl == null && (driverClass == null || driverClass.length() == 0)) 
                    throw new SQLException("url not set");
                
                driver = JdbcUtils.createDriver(driverClassLoader, driverClass);
            
         else 
            if (this.driverClass == null) 
                this.driverClass = driver.getClass().getName();
            
        
    

其中durid自己的mock驱动和clickhouse的负载均衡的驱动，特殊判断了下，其他走的都是class forname.

之后是exception sorter和checker的一些东西，跟主线剧情关系不大，skip.

之后是一些初始化JdbcDataSourceStat,没啥东西。

之后是核心:

  connections = new DruidConnectionHolder[maxActive];  //连接数组
            evictConnections = new DruidConnectionHolder[maxActive]; //销毁的连接数组
            keepAliveConnections = new DruidConnectionHolder[maxActive]; //保持活跃可用的数组

dataSource的连接，都被包装在类DruidConnectionHolder中，之后是一个同步去初始化连接还是异步去初始化的连接，总之，是去初始化 连接的过程:

if (createScheduler != null && asyncInit) 
                for (int i = 0; i < initialSize; ++i) 
                    submitCreateTask(true);
                
             else if (!asyncInit) 
                // init connections
                while (poolingCount < initialSize) 
                    try 
                        PhysicalConnectionInfo pyConnectInfo = createPhysicalConnection();
                        DruidConnectionHolder holder = new DruidConnectionHolder(this, pyConnectInfo);
                        connections[poolingCount++] = holder;
                     catch (SQLException ex) 
                        LOG.error("init datasource error, url: " + this.getUrl(), ex);
                        if (initExceptionThrow) 
                            connectError = ex;
                            break;
                         else 
                            Thread.sleep(3000);
                        
                    
                

                if (poolingCount > 0) 
                    poolingPeak = poolingCount;
                    poolingPeakTime = System.currentTimeMillis();
                
            

初始化的连接个数为连接串里面配置的initialSize.

核心初始化方法com.alibaba.druid.pool.DruidAbstractDataSource#createPhysicalConnection(),在这方法里面，会拿用户名密码，之后执行真正的获取connection:

 public Connection createPhysicalConnection(String url, Properties info) throws SQLException 
        Connection conn;
        if (getProxyFilters().size() == 0) 
            conn = getDriver().connect(url, info);
         else 
            conn = new FilterChainImpl(this).connection_connect(info);
        

        createCountUpdater.incrementAndGet(this);

        return conn;
    

注意，如果配置了filters,则所有操作，都会在操作前执行filter处理链。

 public ConnectionProxy connection_connect(Properties info) throws SQLException 
        if (this.pos < filterSize) 
            return nextFilter()
                    .connection_connect(this, info);
        

        Driver driver = dataSource.getRawDriver();
        String url = dataSource.getRawJdbcUrl();

        Connection nativeConnection = driver.connect(url, info);

        if (nativeConnection == null) 
            return null;
        

        return new ConnectionProxyImpl(dataSource, nativeConnection, info, dataSource.createConnectionId());
    

再回到主流程init方法，connections数组初始化完成之后，
开启额外线程：

     createAndLogThread();  //打印连接信息
            createAndStartCreatorThread(); //创建连接线程
            createAndStartDestroyThread(); //销毁连接线程 

先看注释，具体里面的内容后面单独拉出来讲。

之后：

 initedLatch.await(); //初始化 latch -1
            init = true;  //标记已经初始化完成

            initedTime = new Date(); //时间
            registerMbean(); //为datasource 注册jmx监控指标

最后的最后，如果配置了keepAlive:

            if (keepAlive) 
                // async fill to minIdle
                if (createScheduler != null) 
                    for (int i = 0; i < minIdle; ++i) 
                        submitCreateTask(true);
                    
                 else 
                    this.emptySignal();
                
            

这时候，会根据配置的活跃连接数minIdle,去给datasource的连接，做个保持活跃连接个数，具体后面再说。

连接池使用的核心逻辑

首先，使用数组作为连接的容器，对于真实连接的加入和移除，使用lock就行同步，另外，在加入和移除连接时候，对比生产消费模型，通过lock上的条件，来通知是否可以获取或者加入连接。

 public DruidAbstractDataSource(boolean lockFair)
        lock = new ReentrantLock(lockFair);

        notEmpty = lock.newCondition();  //非空，有连接
        empty = lock.newCondition(); //空的
     

另外，默认的fairlock为false

  public DruidDataSource()
        this(false);
    

    public DruidDataSource(boolean fairLock)
        super(fairLock);

        configFromPropety(System.getProperties());
    

创建连接

在线程com.alibaba.druid.pool.DruidDataSource.CreateConnectionThread中：

 if (emptyWait) 
                        // 必须存在线程等待，才创建连接
                        if (poolingCount >= notEmptyWaitThreadCount //
                                && (!(keepAlive && activeCount + poolingCount < minIdle))
                                && !isFailContinuous()
                        ) 
                            empty.await();
                        

                        // 防止创建超过maxActive数量的连接
                        if (activeCount + poolingCount >= maxActive) 
                            empty.await();
                            continue;
                        
                    

必须存在线程等待获取连接时候，才能创建连接，并且要保持总的连接数，不能超过配置的最大连接。

创建完连接之后，执行notEmpty.signalAll();通知消费者。

获取连接

外层代码：

 @Override
    public DruidPooledConnection getConnection() throws SQLException 
        return getConnection(maxWait);
    

    public DruidPooledConnection getConnection(long maxWaitMillis) throws SQLException 
        init();

        if (filters.size() > 0) 
            FilterChainImpl filterChain = new FilterChainImpl(this);
            return filterChain.dataSource_connect(this, maxWaitMillis);
         else 
            return getConnectionDirect(maxWaitMillis);
        
    

忽略掉filter chain,其实最后执行的还是com.alibaba.druid.pool.DruidDataSource#getConnectionDirect：

方法内部：

   poolableConnection = getConnectionInternal(maxWaitMillis);

• 1 , 连接不足，需要直接去创建新的，跟我们初始化一样
• 2，从connections里面拿

 if (maxWait > 0) 
                    holder = pollLast(nanos);
                 else 
                    holder = takeLast();
                

其中，maxWait默认为-1，配置在init里面：

 String property = properties.getProperty("druid.maxWait");
            if (property != null && property.length() > 0) 
                try 
                    int value = Integer.parseInt(property);
                    this.setMaxWait(value);
                 catch (NumberFormatException e) 
                    LOG.error("illegal property 'druid.maxWait'", e);
                
            

这个用于配置拿连接时候，是否在这个时间上进行等待，默认是否，即一直等到拿到连接为止。

直接看下阻塞拿的过程：

 DruidConnectionHolder takeLast() throws InterruptedException, SQLException 
        try 
            //没连接了
            while (poolingCount == 0) 
                //暗示下创建线程没连接了
                emptySignal(); // send signal to CreateThread create connection

                if (failFast && isFailContinuous()) 
                    throw new DataSourceNotAvailableException(createError);
                

                notEmptyWaitThreadCount++;
                if (notEmptyWaitThreadCount > notEmptyWaitThreadPeak) 
                    notEmptyWaitThreadPeak = notEmptyWaitThreadCount;
                
                try 
                    //傻等着创建或者回收，能给整出来点儿连接
                    notEmpty.await(); // signal by recycle or creator
                 finally 
                    notEmptyWaitThreadCount--;
                
                notEmptyWaitCount++;

                if (!enable) 
                    connectErrorCountUpdater.incrementAndGet(this);
                    if (disableException != null) 
                        throw disableException;
                    

                    throw new DataSourceDisableException();
                
            
         catch (InterruptedException ie) 
            notEmpty.signal(); // propagate to non-interrupted thread
            notEmptySignalCount++;
            throw ie;
        

        //拿数组的最后一个连接
        decrementPoolingCount();
        DruidConnectionHolder last = connections[poolingCount];
        connections[poolingCount] = null;

        return last;
    

连接回收

 protected void createAndStartDestroyThread() 
        destroyTask = new DestroyTask();
	//自定义配置销毁 ，适用于连接数非常多的 情况
        if (destroyScheduler != null) 
            long period = timeBetweenEvictionRunsMillis;
            if (period <= 0) 
                period = 1000;
            
            destroySchedulerFuture = destroyScheduler.scheduleAtFixedRate(destroyTask, period, period,
                                                                          TimeUnit.MILLISECONDS);
            initedLatch.countDown();
            return;
        

        String threadName = "Druid-ConnectionPool-Destroy-" + System.identityHashCode(this);
        //单线程销毁 
        destroyConnectionThread = new DestroyConnectionThread(threadName);
        destroyConnectionThread.start();
    

实际的销毁：

 public class DestroyTask implements Runnable 
        public DestroyTask() 

        

        @Override
        public void run() 
            shrink(true, keepAlive);

            if (isRemoveAbandoned()) 
                removeAbandoned();
            
        

    

最终 执行的还是 shrink方法。

   public void shrink(boolean checkTime, boolean keepAlive) 
        try 
            lock.lockInterruptibly();
         catch (InterruptedException e) 
            return;
        

        boolean needFill = false;
        int evictCount = 0;
        int keepAliveCount = 0;
        int fatalErrorIncrement = fatalErrorCount - fatalErrorCountLastShrink;
        fatalErrorCountLastShrink = fatalErrorCount;
        
        try 
            if (!inited) 
                return;
            

            final int checkCount = poolingCount - minIdle; //需要检测连接的数量
            final long currentTimeMillis = System.currentTimeMillis();
            for (int i = 0; i < poolingCount; ++i)  //检测目前connections数组中的连接
                DruidConnectionHolder connection = connections[i];

                if ((onFatalError || fatalErrorIncrement > 0) && (lastFatalErrorTimeMillis > connection.connectTimeMillis))  
                    keepAliveConnections[keepAliveCount++] = connection;
                    continue;
                

                if (checkTime) 
                    //是否设置了物理连接的超时时间phyTimoutMills。假如设置了该时间，
                    // 判断连接时间存活时间是否已经超过phyTimeoutMills，是则放入evictConnections中
                    if (phyTimeoutMillis > 0) 
                        long phyConnectTimeMillis = currentTimeMillis - connection.connectTimeMillis;
                        if (phyConnectTimeMillis > phyTimeoutMillis) 
                            evictConnections[evictCount++] = connection;
                            continue;
                        
                    

                    long idleMillis = currentTimeMillis - connection.lastActiveTimeMillis;//获取连接空闲时间
                    //如果某条连接空闲时间小于minEvictableIdleTimeMillis，则不用继续检查剩下的连接了
                    if (idleMillis < minEvictableIdleTimeMillis
                            && idleMillis < keepAliveBetweenTimeMillis
                    ) 
                        break;
                    

                    if (idleMillis >= minEvictableIdleTimeMillis) 
                        // check checkTime is silly code
                        //检测检查了几个连接了
                        if (checkTime && i < checkCount) 
                            //超时了
                            evictConnections[evictCount++] = connection;
                            continue;
                         else if (idleMillis > maxEvictableIdleTimeMillis) 
                            //超时了
                            evictConnections[evictCount++] = connection;
                            continue;
                        
                    

                    if (keepAlive && idleMillis >= keepAliveBetweenTimeMillis) 
                        //配置了keepAlive，并且在存活时间内，放到keepAlive数组
                        keepAliveConnections[keepAliveCount++] = connection;
                    
                 else 
                    //不需要检查时间的，直接移除
                    if (i < checkCount) 
                        evictConnections[evictCount++] = connection;
                     else 
                        break;
                    
                
            

            int removeCount = evictCount + keepAliveCount; //移除了几个
            //由于使用connections连接时候，都是取后面的，后面 的是最新的连接，只考虑前面过期就行，所以只需要挪动前面的连接
            if (removeCount > 0) 
                System.arraycopy(connections, removeCount, connections, 0, poolingCount - removeCount);
                Arrays.fill(connections, poolingCount - removeCount, poolingCount, null);
                poolingCount -= removeCount;
            
            keepAliveCheckCount += keepAliveCount;

            if (keepAlive && poolingCount + activeCount < minIdle) 
                //不够核心的活跃连接时候，需要去创建啦
                needFill = true;
            
         finally 
            lock.unlock();
        

        if (evictCount > 0) 
            for (int i = 0; i < evictCount; ++i) 
                //销毁连接
                DruidConnectionHolder item = evictConnections[i];
                Connection connection = item.getConnection();
                JdbcUtils.close(connection);
                destroyCountUpdater.incrementAndGet(this);
            
            Arrays.fill(evictConnections, null);
        

        if (keepAliveCount > 0) 
            // keep order
            for (int i = keepAliveCount - 1; i >= 0; --i) 
                DruidConnectionHolder holer = keepAliveConnections[i];
                Connection connection = holer.getConnection();
                holer.incrementKeepAliveCheckCount();

                boolean validate = false;
                try 
                    this.validateConnection(connection);
                    validate = true;
                 catch (Throwable error) 
                    if (LOG.isDebugEnabled()) 
                        LOG.debug("keepAliveErr", error);
                    
                    // skip
                

                boolean discard = !validate; //没通过validate
                if (validate) 
                    //通过keep alive检查，更新时间
                    holer.lastKeepTimeMillis = System.currentTimeMillis();
                    //这里还会尝试放回connections数组
                    boolean putOk = put(holer, 0L, true);
                    if (!putOk) 
                        //没放入，标记要丢弃了
                        discard = true;
                    
                

                if (discard) 
                    try 
                        connection.close();
                     catch (Exception e) 
                        // skip
                    

                    lock.lock();
                    try 
                        discardCount++;

                        if (activeCount + poolingCount <= minIdle) 
                            //发信号让创建线程去创建
                            emptySignal();
                        
                     finally 
                        lock.unlock();
                    
                
            
            this.getDataSourceStat().addKeepAliveCheckCount(keepAliveCount);
            Arrays.fill(keepAliveConnections, null);
        

        if (needFill) 
            //又要去创建了
            lock.lock();
            try 
                int fillCount = minIdle - (activeCount + poolingCount + createTaskCount);
                for (int i = 0; i < fillCount; ++i) 
                    emptySignal();
                
             finally 
                lock.unlock();
            
         else if (onFatalError || fatalErrorIncrement > 0) 
            lock.lock();
            try 
                emptySignal();
             finally 
                lock.unlock();
            
        
    

工具数组evictConnections,keepAliveConnections 用完即被置空，老工具人了。

一波操作下来，完成了对connections数组的大清洗。

小结

• 只写了核心逻辑，很多validate,checker,filter省略了。
• druid连接池源码里面还有很多好用的工具，比如数据库驱动工具，jdbc工具，解析SQL的语法树，ibatis的支持，wall过滤，多数据源…
• 最新的代码我看还有支持配套ZK的高可用方案，用到的话后期我会继续更新源码解析。