dubbo之Zookeeper注册中心

目前dubbo支持多种注册中心：Zookeeper、Redis、Simple、Multicast、Etcd3。

本编文章是分析使用Zookeeper做为注册中心，dubbo如何整合Zookeeper进行服务注册和订阅服务。

首先dubbo将服务注册到Zookeeper后，目录结构以下所示：（注册接口名：com.bob.dubbo.service.CityDubboService）

在consumer和provider服务启动的时候，去把自身URL格式化成字符串，而后注册到zookeeper相应节点下，做为临时节点，断开链接后，节点删除；consumer启动时，不只会订阅服务，同时也会将本身的URL注册到zookeeper中；

ZookeeperRegistry

ZookeeperRegistry：dubbo与zookeeper交互主要的类，已下结合源码进行分析，先来看

• doSubcribe()

这个方法主要是用于订阅服务，添加监听器，动态监听提供者列表变化：

@Override
    public void doSubscribe(final URL url, final NotifyListener listener) {
        try {
            // 处理全部service层发起的订阅，例如监控中心的订阅
            if (Constants.ANY_VALUE.equals(url.getServiceInterface())) {
                String root = toRootPath();
                ConcurrentMap<NotifyListener, ChildListener> listeners = zkListeners.get(url);
                if (listeners == null) {
                    zkListeners.putIfAbsent(url, new ConcurrentHashMap<>());
                    listeners = zkListeners.get(url);
                }
                ChildListener zkListener = listeners.get(listener);
                if (zkListener == null) {
                    listeners.putIfAbsent(listener, (parentPath, currentChilds) -> {
                        for (String child : currentChilds) {
                            child = URL.decode(child);
                            if (!anyServices.contains(child)) {
                                anyServices.add(child);
                                subscribe(url.setPath(child).addParameters(Constants.INTERFACE_KEY, child,
                                        Constants.CHECK_KEY, String.valueOf(false)), listener);
                            }
                        }
                    });
                    zkListener = listeners.get(listener);
                }
                zkClient.create(root, false);
                List<String> services = zkClient.addChildListener(root, zkListener);
                if (services != null && !services.isEmpty()) {
                    for (String service : services) {
                        service = URL.decode(service);
                        anyServices.add(service);
                        subscribe(url.setPath(service).addParameters(Constants.INTERFACE_KEY, service,
                                Constants.CHECK_KEY, String.valueOf(false)), listener);
                    }
                }
                // 处理指定service层发起的订阅，例如服务消费者的订阅
            } else {
                List<URL> urls = new ArrayList<>();
                // 循环分类数组 , router, configurator, provider
                for (String path : toCategoriesPath(url)) {
                    // 得到 url 对应的监听器集合
                    ConcurrentMap<NotifyListener, ChildListener> listeners = zkListeners.get(url);
                    if (listeners == null) {// 不存在，进行建立
                        zkListeners.putIfAbsent(url, new ConcurrentHashMap<>());
                        listeners = zkListeners.get(url);
                    }
                    // 得到 ChildListener 对象
                    ChildListener zkListener = listeners.get(listener);
                    if (zkListener == null) {//  不存在子目录的监听器，进行建立 ChildListener 对象
                        // 订阅父级目录, 当有子节点发生变化时，触发此回调函数，回调listener中的notify()方法
                        listeners.putIfAbsent(listener, (parentPath, currentChilds) -> ZookeeperRegistry.this.notify(url, listener, toUrlsWithEmpty(url, parentPath, currentChilds)));
                        zkListener = listeners.get(listener);
                    }
                    建立Type节点，此节点为持久节点
                    zkClient.create(path, false);
                    // 向 Zookeeper ，PATH 节点，发起订阅,返回此节点下的全部子元素 path : /根节点/接口全名/providers, 好比 ： /dubbo/com.bob.service.CityService/providers
                    List<String> children = zkClient.addChildListener(path, zkListener);
                    if (children != null) {
                        urls.addAll(toUrlsWithEmpty(url, path, children));
                    }
                }
                // 首次全量数据获取完成时，调用 `#notify(...)` 方法，回调 NotifyListener, 在这一步从链接Provider,实例化Invoker
                notify(url, listener, urls);
            }
        } catch (Throwable e) {
            throw new RpcException("Failed to subscribe " + url + " to zookeeper " + getUrl() + ", cause: " + e.getMessage(), e);
        }
    }

• register()

ZookeeperRegistry父类FailbackRegistry中的方法，用于将服务注册到zookeeper，具体代码以下：

@Override
    public void register(URL url) {
        // 调用父类AbstractRegistry中的register()方法，将url存储到注册集合中
        super.register(url);
        // 若是以前这个url注册失败，则会从注册失败集合中删除
        removeFailedRegistered(url);
        removeFailedUnregistered(url);
        try {
            // 像注册中心发送注册请求
            doRegister(url);
        } catch (Exception e) {
            Throwable t = e;

            // If the startup detection is opened, the Exception is thrown directly.
            boolean check = getUrl().getParameter(Constants.CHECK_KEY, true)
                    && url.getParameter(Constants.CHECK_KEY, true)
                    && !Constants.CONSUMER_PROTOCOL.equals(url.getProtocol());
            boolean skipFailback = t instanceof SkipFailbackWrapperException;
            if (check || skipFailback) {
                if (skipFailback) {
                    t = t.getCause();
                }
                throw new IllegalStateException("Failed to register " + url + " to registry " + getUrl().getAddress() + ", cause: " + t.getMessage(), t);
            } else {
                logger.error("Failed to register " + url + ", waiting for retry, cause: " + t.getMessage(), t);
            }

            // 将url存入注册失败集合中，进行重试try()
            addFailedRegistered(url);
        }
    }

• doRegister()

ZookeeperRegistry类中的方法

@Override
    public void doRegister(URL url) {
        try {
            // 经过zookeeper客户端向注册中心发送服务注册请求，在zookeeper下建立服务对应的节点
            zkClient.create(toUrlPath(url), url.getParameter(Constants.DYNAMIC_KEY, true));
        } catch (Throwable e) {
            throw new RpcException("Failed to register " + url + " to zookeeper " + getUrl() + ", cause: " + e.getMessage(), e);
        }
    }

在介绍注册registry()方法的时候，解析到了FailbackRegistry类，接下来我们来分析一下这个类的做用：

FailbackRegistry

这个类是ZookeeperRegistry的父类，经过分析该类的结构，主要是用于服务的注册、订阅、重试，而服务具体的注册、订阅又在ZookeeperRegistry子类进行了实现，如今咱们来分析重试这个功能，服务暴露和订阅的配置文件中通常会设置重试这个属性，以下所示：

<dubbo:service interface="com.fy.view.service.ProductManageService" ref="productManageServiceImpl" retries="2"/>

上面是一个服务暴露的示例，设置了retries属性，表示重试的次数。接下来我们就以注册重试进行分析（服务订阅是一样的原理）：在注册registry()方法中(代码上面已提供)，在异常catch{}代码块中有一个addFailedRegistered(url)方法，这个就是将注册失败的url添加到集合中，并建立一个重试的任务FailedRegisteredTask(url, this)，代码以下：

private void addFailedRegistered(URL url) {
        // 先从集合中获取，若是存在，直接返回
        FailedRegisteredTask oldOne = failedRegistered.get(url);
        if (oldOne != null) {
            return;
        }
        // 本地集合不存在，则建立重试定时任务，默认每隔5s执行
        FailedRegisteredTask newTask = new FailedRegisteredTask(url, this);
        oldOne = failedRegistered.putIfAbsent(url, newTask);
        if (oldOne == null) {
            // 将定时任务放置在HashedWheelTimer这个处理定时任务的容器，(HashedWheelTimer执行原理，能够自行查找资料，这里就不介绍)
            retryTimer.newTimeout(newTask, retryPeriod, TimeUnit.MILLISECONDS);
        }
    }

我们下来看FailedRegisteredTask这个定时任务，有哪些东西，FailedRegisteredTask是AbstractRetryTask的子类，在执行new FailedRegisteredTask(url, this)代码时，其实调用的是父类构造函数，其中retryTimes表示重试的次数，在没有配置的状况下，默认重试三次：

AbstractRetryTask(URL url, FailbackRegistry registry, String taskName) {
        if (url == null || StringUtils.isBlank(taskName)) {
            throw new IllegalArgumentException();
        }
        this.url = url;
        this.registry = registry;
        this.taskName = taskName;
        cancel = false;
        this.retryPeriod = url.getParameter(Constants.REGISTRY_RETRY_PERIOD_KEY, Constants.DEFAULT_REGISTRY_RETRY_PERIOD);
        // 重试次数，默认状况下重试三次
        this.retryTimes = url.getParameter(Constants.REGISTRY_RETRY_TIMES_KEY, Constants.DEFAULT_REGISTRY_RETRY_TIMES);
    }

在AbstractRetryTask类中有一个run()方法，在run()方法会根据XML配置文件中的retries属性值进行比较来进行重试，若是没有达到重试次数，则会调用doRetry(url, registry, timeout)，而这个方法又在子类具体实现，这里我以注册FailedRegisteredTask举例：

@Override
    public void run(Timeout timeout) throws Exception {
        if (timeout.isCancelled() || timeout.timer().isStop() || isCancel()) {
            // other thread cancel this timeout or stop the timer.
            return;
        }
        // 重试次数与设置的retries进行比较，超过则不在进行重试
        if (times > retryTimes) {
            // reach the most times of retry.
            logger.warn("Final failed to execute task " + taskName + ", url: " + url + ", retry " + retryTimes + " times.");
            return;
        }
        if (logger.isInfoEnabled()) {
            logger.info(taskName + " : " + url);
        }
        try {
            // 调用子类实现，进行重试
            doRetry(url, registry, timeout);
        } catch (Throwable t) { // Ignore all the exceptions and wait for the next retry
            logger.warn("Failed to execute task " + taskName + ", url: " + url + ", waiting for again, cause:" + t.getMessage(), t);
            // reput this task when catch exception.
            reput(timeout, retryPeriod);
        }
    }

在子类FailedRegisteredTask中doRetry()方法具体实现：

public final class FailedRegisteredTask extends AbstractRetryTask {

    private static final String NAME = "retry register";

    public FailedRegisteredTask(URL url, FailbackRegistry registry) {
        super(url, registry, NAME);
    }

    @Override
    protected void doRetry(URL url, FailbackRegistry registry, Timeout timeout) {
        // 调用ZookeeperRegistry类中的doRegister()方法进行注册
        registry.doRegister(url);
        registry.removeFailedRegisteredTask(url);
    }
}

分析到这里，有个疑问：重试任务已经封装了，任务何时去执行，怎么执行的？其实在上面我们就分析到过，就是使用了HashedWheelTimer，这个类是在ZookeeperRegistry类初始化的时候就会去初始化：

public ZookeeperRegistry(URL url, ZookeeperTransporter zookeeperTransporter) {
        // 这个地方进行初始化的：初始化父类FailbackRegistry
        super(url);
        if (url.isAnyHost()) {
            throw new IllegalStateException("registry address == null");
        }
        String group = url.getParameter(Constants.GROUP_KEY, DEFAULT_ROOT);
        if (!group.startsWith(Constants.PATH_SEPARATOR)) {
            group = Constants.PATH_SEPARATOR + group;
        }
        this.root = group;
        zkClient = zookeeperTransporter.connect(url);
        zkClient.addStateListener(state -> {
            if (state == StateListener.RECONNECTED) {
                try {
                    recover();
                } catch (Exception e) {
                    logger.error(e.getMessage(), e);
                }
            }
        });
    }

public FailbackRegistry(URL url) {
        super(url);
        this.retryPeriod = url.getParameter(Constants.REGISTRY_RETRY_PERIOD_KEY, Constants.DEFAULT_REGISTRY_RETRY_PERIOD);

        // 建立HashedWheelTimer对象
        retryTimer = new HashedWheelTimer(new NamedThreadFactory("DubboRegistryRetryTimer", true), retryPeriod, TimeUnit.MILLISECONDS, 128);
    }

而后在addFailedRegistered()方法中有retryTimer.newTimeout(newTask, retryPeriod, TimeUnit.MILLISECONDS);这样的一条代码，这个就是执行任务的开始点：

@Override
    public Timeout newTimeout(TimerTask task, long delay, TimeUnit unit) {
        if (task == null) {
            throw new NullPointerException("task");
        }
        if (unit == null) {
            throw new NullPointerException("unit");
        }

        long pendingTimeoutsCount = pendingTimeouts.incrementAndGet();

        if (maxPendingTimeouts > 0 && pendingTimeoutsCount > maxPendingTimeouts) {
            pendingTimeouts.decrementAndGet();
            throw new RejectedExecutionException("Number of pending timeouts ("
                    + pendingTimeoutsCount + ") is greater than or equal to maximum allowed pending "
                    + "timeouts (" + maxPendingTimeouts + ")");
        }
        // 开启轮询任务
        start();

        // Add the timeout to the timeout queue which will be processed on the next tick.
        // During processing all the queued HashedWheelTimeouts will be added to the correct HashedWheelBucket.
        long deadline = System.nanoTime() + unit.toNanos(delay) - startTime;

        // Guard against overflow.
        if (delay > 0 && deadline < 0) {
            deadline = Long.MAX_VALUE;
        }
        HashedWheelTimeout timeout = new HashedWheelTimeout(this, task, deadline);
        timeouts.add(timeout);
        return timeout;
    }

调用start()方法时，开启一个线程work去轮询存储到HashedWheelTimer容器的任务，而后调用任务中的run()方法，

public void start() {
        switch (WORKER_STATE_UPDATER.get(this)) {
            case WORKER_STATE_INIT:
                if (WORKER_STATE_UPDATER.compareAndSet(this, WORKER_STATE_INIT, WORKER_STATE_STARTED)) {
                    // 开启work线程，执行work线程中的run()方法
                    workerThread.start();
                }
                break;
            case WORKER_STATE_STARTED:
                break;
            case WORKER_STATE_SHUTDOWN:
                throw new IllegalStateException("cannot be started once stopped");
            default:
                throw new Error("Invalid WorkerState");
        }

        // Wait until the startTime is initialized by the worker.
        while (startTime == 0) {
            try {
                startTimeInitialized.await();
            } catch (InterruptedException ignore) {
                // Ignore - it will be ready very soon.
            }
        }
    }

@Override
        public void run() {
            // Initialize the startTime.
            startTime = System.nanoTime();
            if (startTime == 0) {
                // We use 0 as an indicator for the uninitialized value here, so make sure it's not 0 when initialized.
                startTime = 1;
            }

            // Notify the other threads waiting for the initialization at start().
            startTimeInitialized.countDown();

            do {
                final long deadline = waitForNextTick();
                if (deadline > 0) {
                    int idx = (int) (tick & mask);
                    processCancelledTasks();
                    HashedWheelBucket bucket =
                            wheel[idx];
                    transferTimeoutsToBuckets();
                    // 执行重试任务
                    bucket.expireTimeouts(deadline);
                    tick++;
                }
            } while (WORKER_STATE_UPDATER.get(HashedWheelTimer.this) == WORKER_STATE_STARTED);

            // Fill the unprocessedTimeouts so we can return them from stop() method.
            for (HashedWheelBucket bucket : wheel) {
                bucket.clearTimeouts(unprocessedTimeouts);
            }
            for (; ; ) {
                HashedWheelTimeout timeout = timeouts.poll();
                if (timeout == null) {
                    break;
                }
                if (!timeout.isCancelled()) {
                    unprocessedTimeouts.add(timeout);
                }
            }
            processCancelledTasks();
        }

void expireTimeouts(long deadline) {
            HashedWheelTimeout timeout = head;

            // process all timeouts
            while (timeout != null) {
                // 轮询获取重试任务
                HashedWheelTimeout next = timeout.next;
                if (timeout.remainingRounds <= 0) {
                    next = remove(timeout);
                    if (timeout.deadline <= deadline) {
                        // 执行重试任务
                        timeout.expire();
                    } else {
                        // The timeout was placed into a wrong slot. This should never happen.
                        throw new IllegalStateException(String.format(
                                "timeout.deadline (%d) > deadline (%d)", timeout.deadline, deadline));
                    }
                } else if (timeout.isCancelled()) {
                    next = remove(timeout);
                } else {
                    timeout.remainingRounds--;
                }
                timeout = next;
            }
        }

public void expire() {
            if (!compareAndSetState(ST_INIT, ST_EXPIRED)) {
                return;
            }

            try {
                // 调用任务中的run()方法，（如：AbstractRetryTask任务中的run()方法，在去调用子类FailedRegisteredTask中的doRetry()方法进行重试注册）
                task.run(this);
            } catch (Throwable t) {
                if (logger.isWarnEnabled()) {
                    logger.warn("An exception was thrown by " + TimerTask.class.getSimpleName() + '.', t);
                }
            }
        }

在上面对于HashedWheelTimer的具体实现原理，并无进行详细的进行分析，若是想了解的和学习的话，能够自行查找资料。