dubbo重试和超时机制

　　dubbo 超时机制以及负载均衡、重试机制都是针对客户端进行的。

1. dubbo 重试机制

　　dubbo 重试机制针对不同的Invoker。主要的集群Invoker 有如下：

 默认的集群Invoker是FailoverClusterInvoker。这里有重试机制。其默认的重试次数是2次（调用1次，重试2次，所以总共尝试是3次）。

package org.apache.dubbo.rpc.cluster.support;

import org.apache.dubbo.common.Version;
import org.apache.dubbo.common.logger.Logger;
import org.apache.dubbo.common.logger.LoggerFactory;
import org.apache.dubbo.common.utils.NetUtils;
import org.apache.dubbo.rpc.Invocation;
import org.apache.dubbo.rpc.Invoker;
import org.apache.dubbo.rpc.Result;
import org.apache.dubbo.rpc.RpcContext;
import org.apache.dubbo.rpc.RpcException;
import org.apache.dubbo.rpc.cluster.Directory;
import org.apache.dubbo.rpc.cluster.LoadBalance;
import org.apache.dubbo.rpc.support.RpcUtils;

import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Set;

import static org.apache.dubbo.rpc.cluster.Constants.DEFAULT_RETRIES;
import static org.apache.dubbo.rpc.cluster.Constants.RETRIES_KEY;

/**
 * When invoke fails, log the initial error and retry other invokers (retry n times, which means at most n different invokers will be invoked)
 * Note that retry causes latency.
 * <p>
 * <a href="http://en.wikipedia.org/wiki/Failover">Failover</a>
 *
 */
public class FailoverClusterInvoker<T> extends AbstractClusterInvoker<T> {

    private static final Logger logger = LoggerFactory.getLogger(FailoverClusterInvoker.class);

    public FailoverClusterInvoker(Directory<T> directory) {
        super(directory);
    }

    @Override
    @SuppressWarnings({"unchecked", "rawtypes"})
    public Result doInvoke(Invocation invocation, final List<Invoker<T>> invokers, LoadBalance loadbalance) throws RpcException {
        List<Invoker<T>> copyInvokers = invokers;
        checkInvokers(copyInvokers, invocation);
        String methodName = RpcUtils.getMethodName(invocation);
        int len = getUrl().getMethodParameter(methodName, RETRIES_KEY, DEFAULT_RETRIES) + 1;
        if (len <= 0) {
            len = 1;
        }
        // retry loop.
        RpcException le = null; // last exception.
        List<Invoker<T>> invoked = new ArrayList<Invoker<T>>(copyInvokers.size()); // invoked invokers.
        Set<String> providers = new HashSet<String>(len);
        for (int i = 0; i < len; i++) {
            //Reselect before retry to avoid a change of candidate `invokers`.
            //NOTE: if `invokers` changed, then `invoked` also lose accuracy.
            if (i > 0) {
                checkWhetherDestroyed();
                copyInvokers = list(invocation);
                // check again
                checkInvokers(copyInvokers, invocation);
            }
            Invoker<T> invoker = select(loadbalance, invocation, copyInvokers, invoked);
            invoked.add(invoker);
            RpcContext.getContext().setInvokers((List) invoked);
            try {
                Result result = invoker.invoke(invocation);
                if (le != null && logger.isWarnEnabled()) {
                    logger.warn("Although retry the method " + methodName
                            + " in the service " + getInterface().getName()
                            + " was successful by the provider " + invoker.getUrl().getAddress()
                            + ", but there have been failed providers " + providers
                            + " (" + providers.size() + "/" + copyInvokers.size()
                            + ") from the registry " + directory.getUrl().getAddress()
                            + " on the consumer " + NetUtils.getLocalHost()
                            + " using the dubbo version " + Version.getVersion() + ". Last error is: "
                            + le.getMessage(), le);
                }
                return result;
            } catch (RpcException e) {
                if (e.isBiz()) { // biz exception.
                    throw e;
                }
                le = e;
            } catch (Throwable e) {
                le = new RpcException(e.getMessage(), e);
            } finally {
                providers.add(invoker.getUrl().getAddress());
            }
        }
        throw new RpcException(le.getCode(), "Failed to invoke the method "
                + methodName + " in the service " + getInterface().getName()
                + ". Tried " + len + " times of the providers " + providers
                + " (" + providers.size() + "/" + copyInvokers.size()
                + ") from the registry " + directory.getUrl().getAddress()
                + " on the consumer " + NetUtils.getLocalHost() + " using the dubbo version "
                + Version.getVersion() + ". Last error is: "
                + le.getMessage(), le.getCause() != null ? le.getCause() : le);
    }

}

　　如上代码可以看到负载均衡和重试机制。会重试三次，然后进行负载均衡选择服务器之后进行RPC调用。

　　如果正常则返回结果，如果重试都失败之后就抛出RPC异常。

 

还有一些其他的Invoker 调用策略，比如：

- failsafe， 这个策略默认是返回一个空值，失败不做任何异常处理。

public class FailsafeClusterInvoker<T> extends AbstractClusterInvoker<T> {
    private static final Logger logger = LoggerFactory.getLogger(FailsafeClusterInvoker.class);

    public FailsafeClusterInvoker(Directory<T> directory) {
        super(directory);
    }

    @Override
    public Result doInvoke(Invocation invocation, List<Invoker<T>> invokers, LoadBalance loadbalance) throws RpcException {
        try {
            checkInvokers(invokers, invocation);
            Invoker<T> invoker = select(loadbalance, invocation, invokers, null);
            return invoker.invoke(invocation);
        } catch (Throwable e) {
            logger.error("Failsafe ignore exception: " + e.getMessage(), e);
            return AsyncRpcResult.newDefaultAsyncResult(null, null, invocation); // ignore
        }
    }
}

- failfast， 这个策略不进行重试，会直接抛出异常

public class FailfastClusterInvoker<T> extends AbstractClusterInvoker<T> {

    public FailfastClusterInvoker(Directory<T> directory) {
        super(directory);
    }

    @Override
    public Result doInvoke(Invocation invocation, List<Invoker<T>> invokers, LoadBalance loadbalance) throws RpcException {
        checkInvokers(invokers, invocation);
        Invoker<T> invoker = select(loadbalance, invocation, invokers, null);
        try {
            return invoker.invoke(invocation);
        } catch (Throwable e) {
            if (e instanceof RpcException && ((RpcException) e).isBiz()) { // biz exception.
                throw (RpcException) e;
            }
            throw new RpcException(e instanceof RpcException ? ((RpcException) e).getCode() : 0,
                    "Failfast invoke providers " + invoker.getUrl() + " " + loadbalance.getClass().getSimpleName()
                            + " select from all providers " + invokers + " for service " + getInterface().getName()
                            + " method " + invocation.getMethodName() + " on consumer " + NetUtils.getLocalHost()
                            + " use dubbo version " + Version.getVersion()
                            + ", but no luck to perform the invocation. Last error is: " + e.getMessage(),
                    e.getCause() != null ? e.getCause() : e);
        }
    }
}

 

重试次数可以通过修改全局配置，也可以针对单个服务进行修改：

(1) 全局配置

dubbo:
  consumer:
    retries: 3

(2) 单个配置

    @Reference(version = "1.0.0", retries = 5)
    private UserService userService;

2. dubbo 负载均衡

dubbo的负载均衡选择器如下：

(1) Random - 随机算法,(根据权重进行随机) 默认算法

(2) RoundRobin - 轮询(基于权重)负载均衡算法

(3) leastactive - 最近最少活跃

(4) consistenthash - 一致性hash， 计算参数的hash， 根据hash 进行选择

3. 超时机制

　　参考：https://juejin.cn/post/6887553443880255501

　　dubbo 超时一般也是针对消费者端。消费者端的超时时间默认是1000 ms，可以通过配置文件进行修改。下面研究其超时实现：

(1)  dubbo客户端服务调用会调用到如下方法： org.apache.dubbo.rpc.protocol.AsyncToSyncInvoker#invoke：

    public Result invoke(Invocation invocation) throws RpcException {
        Result asyncResult = invoker.invoke(invocation);

        try {
            if (InvokeMode.SYNC == ((RpcInvocation) invocation).getInvokeMode()) {
                asyncResult.get(Integer.MAX_VALUE, TimeUnit.MILLISECONDS);
            }
        } catch (InterruptedException e) {
            throw new RpcException("Interrupted unexpectedly while waiting for remoting result to return!  method: " + invocation.getMethodName() + ", provider: " + getUrl() + ", cause: " + e.getMessage(), e);
        } catch (ExecutionException e) {
            Throwable t = e.getCause();
            if (t instanceof TimeoutException) {
                throw new RpcException(RpcException.TIMEOUT_EXCEPTION, "Invoke remote method timeout. method: " + invocation.getMethodName() + ", provider: " + getUrl() + ", cause: " + e.getMessage(), e);
            } else if (t instanceof RemotingException) {
                throw new RpcException(RpcException.NETWORK_EXCEPTION, "Failed to invoke remote method: " + invocation.getMethodName() + ", provider: " + getUrl() + ", cause: " + e.getMessage(), e);
            }
        } catch (Throwable e) {
            throw new RpcException(e.getMessage(), e);
        }
        return asyncResult;
    }

这里可以看出： 如果是同步模式的话，需要调用asyncResult.get 获取等待结果。如果是异步模式的话直接返回asyncResult， 不等待返回结果。所以同步是通过java.util.concurrent.Future#get(long, java.util.concurrent.TimeUnit) 同步阻塞等待结果实现的。 内部会判断如果是超时异常，会抛出异常。

get 后面会调用到： java.util.concurrent.CompletableFuture#reportGet 然后抛出异常 ExecutionException

    private static <T> T reportGet(Object r)
        throws InterruptedException, ExecutionException {
        if (r == null) // by convention below, null means interrupted
            throw new InterruptedException();
        if (r instanceof AltResult) {
            Throwable x, cause;
            if ((x = ((AltResult)r).ex) == null)
                return null;
            if (x instanceof CancellationException)
                throw (CancellationException)x;
            if ((x instanceof CompletionException) &&
                (cause = x.getCause()) != null)
                x = cause;
            throw new ExecutionException(x);
        }
        @SuppressWarnings("unchecked") T t = (T) r;
        return t;
    }

(2) 下面研究其生成AltResult 的过程

 　　这里可以理解为后面有个定时任务(HashedWheelTimer， 哈希轮定时器)来对发送之后的请求进行判断。如果超过超时时间的设置就设置响应结果为超时，这样前面在get 调用到 reportGet 的时候可以获取到相关的超时异常。

 org.apache.dubbo.rpc.protocol.AsyncToSyncInvoker#invoke 经过一系列调用会调用到  org.apache.dubbo.remoting.exchange.support.header.HeaderExchangeChannel#request(java.lang.Object, int)

    public CompletableFuture<Object> request(Object request, int timeout) throws RemotingException {
        if (closed) {
            throw new RemotingException(this.getLocalAddress(), null, "Failed to send request " + request + ", cause: The channel " + this + " is closed!");
        }
        // create request.
        Request req = new Request();
        req.setVersion(Version.getProtocolVersion());
        req.setTwoWay(true);
        req.setData(request);
        DefaultFuture future = DefaultFuture.newFuture(channel, req, timeout);
        try {
            channel.send(req);
        } catch (RemotingException e) {
            future.cancel();
            throw e;
        }
        return future;
    }

1》org.apache.dubbo.remoting.exchange.support.DefaultFuture#newFuture

    public static DefaultFuture newFuture(Channel channel, Request request, int timeout) {
        final DefaultFuture future = new DefaultFuture(channel, request, timeout);
        // timeout check
        timeoutCheck(future);
        return future;
    }

    private static void timeoutCheck(DefaultFuture future) {
        TimeoutCheckTask task = new TimeoutCheckTask(future.getId());
        future.timeoutCheckTask = TIME_OUT_TIMER.newTimeout(task, future.getTimeout(), TimeUnit.MILLISECONDS);
    }

2》org.apache.dubbo.common.timer.HashedWheelTimer#newTimeout 这里创建超时检测任务并且添加到timeouts 队列中

    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 + ")");
        }

        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;
    }

3》接下来就交给org.apache.dubbo.common.timer.HashedWheelTimer.Worker#run 方法定时跑任务

        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();
                    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();
        }

 然后会调用到： org.apache.dubbo.common.timer.HashedWheelTimer.HashedWheelBucket#expireTimeouts

        /**
         * Expire all {@link HashedWheelTimeout}s for the given {@code deadline}.
         */
        void expireTimeouts(long deadline) {
            HashedWheelTimeout timeout = head;

            // process all timeouts
            while (timeout != null) {
                HashedWheelTimeout next = timeout.next;
                if (timeout.remainingRounds <= 0) {
                    next = remove(timeout);
                    if (timeout.deadline <= deadline) {
                        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;
            }
        }

继续调用调用到：org.apache.dubbo.common.timer.HashedWheelTimer.HashedWheelTimeout#expire

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

            try {
                task.run(this);
            } catch (Throwable t) {
                if (logger.isWarnEnabled()) {
                    logger.warn("An exception was thrown by " + TimerTask.class.getSimpleName() + '.', t);
                }
            }
        }

 继续调用：org.apache.dubbo.remoting.exchange.support.DefaultFuture.TimeoutCheckTask#run

        @Override
        public void run(Timeout timeout) {
            DefaultFuture future = DefaultFuture.getFuture(requestID);
            if (future == null || future.isDone()) {
                return;
            }
            // create exception response.
            Response timeoutResponse = new Response(future.getId());
            // set timeout status.
            timeoutResponse.setStatus(future.isSent() ? Response.SERVER_TIMEOUT : Response.CLIENT_TIMEOUT);
            timeoutResponse.setErrorMessage(future.getTimeoutMessage(true));
            // handle response.
            DefaultFuture.received(future.getChannel(), timeoutResponse, true);

        }

然后调用 org.apache.dubbo.remoting.exchange.support.DefaultFuture#received(org.apache.dubbo.remoting.Channel, org.apache.dubbo.remoting.exchange.Response, boolean) 设置结果为超时结果。 这里的res.getStatus 为SERVER_TIMEOUT。 也就是请求已经发出去，但是在指定的超时时间内没有响应结果。如果在指定时间内，请求还没有发出去则认为是客户端超时。

    public static void received(Channel channel, Response response, boolean timeout) {
        try {
            DefaultFuture future = FUTURES.remove(response.getId());
            if (future != null) {
                Timeout t = future.timeoutCheckTask;
                if (!timeout) {
                    // decrease Time
                    t.cancel();
                }
                future.doReceived(response);
            } else {
                logger.warn("The timeout response finally returned at "
                        + (new SimpleDateFormat("yyyy-MM-dd HH:mm:ss.SSS").format(new Date()))
                        + ", response " + response
                        + (channel == null ? "" : ", channel: " + channel.getLocalAddress()
                        + " -> " + channel.getRemoteAddress()));
            }
        } finally {
            CHANNELS.remove(response.getId());
        }
    }

    private void doReceived(Response res) {
        if (res == null) {
            throw new IllegalStateException("response cannot be null");
        }
        if (res.getStatus() == Response.OK) {
            this.complete(res.getResult());
        } else if (res.getStatus() == Response.CLIENT_TIMEOUT || res.getStatus() == Response.SERVER_TIMEOUT) {
            this.completeExceptionally(new TimeoutException(res.getStatus() == Response.SERVER_TIMEOUT, channel, res.getErrorMessage()));
        } else {
            this.completeExceptionally(new RemotingException(channel, res.getErrorMessage()));
        }
    }

然后调用 java.util.concurrent.CompletableFuture#completeExceptionally 设置结果为：AltResult， 并且记录异常为超时异常。这样在前面的get 阻塞获取结果的时候就可以获取到结果。

    public boolean completeExceptionally(Throwable ex) {
        if (ex == null) throw new NullPointerException();
        boolean triggered = internalComplete(new AltResult(ex));
        postComplete();
        return triggered;
    }

 

补充：重试次数、是否异步等都可以单独配置。

这里需要注意，如果配置不生效，可能是单个服务里面配置了多个Reference。 注解生成的代理对象是单例，所以导致不生效。 最好是将dubbo 接口集中统一管理。

    @Reference(version = "1.0.0", retries = 5, async = true)
    private UserService userService;

查看注解如下：

@Documented
@Retention(RetentionPolicy.RUNTIME)
@Target({ElementType.FIELD, ElementType.METHOD, ElementType.ANNOTATION_TYPE})
public @interface Reference {
    /**
     * Interface class, default value is void.class
     */
    Class<?> interfaceClass() default void.class;

    /**
     * Interface class name, default value is empty string
     */
    String interfaceName() default "";

    /**
     * Service version, default value is empty string
     */
    String version() default "";

    /**
     * Service group, default value is empty string
     */
    String group() default "";

    /**
     * Service target URL for direct invocation, if this is specified, then registry center takes no effect.
     */
    String url() default "";

    /**
     * Client transport type, default value is "netty"
     */
    String client() default "";

    /**
     * Whether to enable generic invocation, default value is false
     */
    boolean generic() default false;

    /**
     * When enable, prefer to call local service in the same JVM if it's present, default value is true
     */
    boolean injvm() default true;

    /**
     * Check if service provider is available during boot up, default value is true
     */
    boolean check() default true;

    /**
     * Whether eager initialize the reference bean when all properties are set, default value is false
     */
    boolean init() default false;

    /**
     * Whether to make connection when the client is created, the default value is false
     */
    boolean lazy() default false;

    /**
     * Export an stub service for event dispatch, default value is false.
     *
     * @see org.apache.dubbo.rpc.Constants#STUB_EVENT_METHODS_KEY
     */
    boolean stubevent() default false;

    /**
     * Whether to reconnect if connection is lost, if not specify, reconnect is enabled by default, and the interval
     * for retry connecting is 2000 ms
     *
     * @see org.apache.dubbo.remoting.Constants#DEFAULT_RECONNECT_PERIOD
     */
    String reconnect() default "";

    /**
     * Whether to stick to the same node in the cluster, the default value is false
     *
     * @see Constants#DEFAULT_CLUSTER_STICKY
     */
    boolean sticky() default false;

    /**
     * How the proxy is generated, legal values include: jdk, javassist
     */
    String proxy() default "";

    /**
     * Service stub name, use interface name + Local if not set
     */
    String stub() default "";

    /**
     * Cluster strategy, legal values include: failover, failfast, failsafe, failback, forking
     */
    String cluster() default "";

    /**
     * Maximum connections service provider can accept, default value is 0 - connection is shared
     */
    int connections() default 0;

    /**
     * The callback instance limit peer connection
     *
     * @see org.apache.dubbo.rpc.Constants#DEFAULT_CALLBACK_INSTANCES
     */
    int callbacks() default 0;

    /**
     * Callback method name when connected, default value is empty string
     */
    String onconnect() default "";

    /**
     * Callback method name when disconnected, default value is empty string
     */
    String ondisconnect() default "";

    /**
     * Service owner, default value is empty string
     */
    String owner() default "";

    /**
     * Service layer, default value is empty string
     */
    String layer() default "";

    /**
     * Service invocation retry times
     *
     * @see Constants#DEFAULT_RETRIES
     */
    int retries() default 2;

    /**
     * Load balance strategy, legal values include: random, roundrobin, leastactive
     *
     * @see Constants#DEFAULT_LOADBALANCE
     */
    String loadbalance() default "";

    /**
     * Whether to enable async invocation, default value is false
     */
    boolean async() default false;

    /**
     * Maximum active requests allowed, default value is 0
     */
    int actives() default 0;

    /**
     * Whether the async request has already been sent, the default value is false
     */
    boolean sent() default false;

    /**
     * Service mock name, use interface name + Mock if not set
     */
    String mock() default "";

    /**
     * Whether to use JSR303 validation, legal values are: true, false
     */
    String validation() default "";

    /**
     * Timeout value for service invocation, default value is 0
     */
    int timeout() default 0;

    /**
     * Specify cache implementation for service invocation, legal values include: lru, threadlocal, jcache
     */
    String cache() default "";

    /**
     * Filters for service invocation
     *
     * @see Filter
     */
    String[] filter() default {};

    /**
     * Listeners for service exporting and unexporting
     *
     * @see ExporterListener
     */
    String[] listener() default {};

    /**
     * Customized parameter key-value pair, for example: {key1, value1, key2, value2}
     */
    String[] parameters() default {};

    /**
     * Application associated name
     */
    String application() default "";

    /**
     * Module associated name
     */
    String module() default "";

    /**
     * Consumer associated name
     */
    String consumer() default "";

    /**
     * Monitor associated name
     */
    String monitor() default "";

    /**
     * Registry associated name
     */
    String[] registry() default {};

    /**
     * The communication protocol of Dubbo Service
     *
     * @return the default value is ""
     * @since 2.6.6
     */
    String protocol() default "";

    /**
     * Service tag name
     */
    String tag() default "";

    /**
     * methods support
     *
     * @return
     */
    Method[] methods() default {};

    /**
     * The id
     *
     * @return default value is empty
     * @since 2.7.3
     */
    String id() default "";
}