Apache kafka 春云侦探与齐普金与卡夫卡

我在用spring cloud的侦探和zipkin以及kafka。 这是我的pom.xml配置

    <dependency>
        <groupId>org.springframework.cloud</groupId>
        <artifactId>spring-cloud-starter-zipkin</artifactId>
        <version>2.0.0.RC1</version>
    </dependency>

    <dependency>
        <groupId>org.springframework.kafka</groupId>
        <artifactId>spring-kafka</artifactId>
    </dependency>

---

然后，我启动SpringBoot应用程序，尝试通过浏览器访问，但我找不到SpringCloudSendSleuth to kafka。因此，我尝试调试zipkin2.reporter.AsyncReporter，并发现每当[flush]方法返回第265行时，即“if（！bundler.isReady（）&&！closed.get（））返回”，代码如下所示

 void flush(BufferNextMessage<S> bundler) {
  if (closed.get()) throw new IllegalStateException("closed");

  pending.drainTo(bundler, bundler.remainingNanos());

  // record after flushing reduces the amount of gauge events vs on doing this on report
  metrics.updateQueuedSpans(pending.count);
  metrics.updateQueuedBytes(pending.sizeInBytes);

  // loop around if we are running, and the bundle isn't full
  // if we are closed, try to send what's pending
  if (!bundler.isReady() && !closed.get()) return;

  // Signal that we are about to send a message of a known size in bytes
  metrics.incrementMessages();
  metrics.incrementMessageBytes(bundler.sizeInBytes());

void flush（BufferNextMessage绑定器）{
如果（closed.get（））抛出新的非法状态异常（“closed”）；
pending.drainTo（bundler，bundler.remainingNanos（））；
//与在报告中执行此操作相比，冲洗后记录减少了仪表事件的数量
metrics.updateQueuedSpans（pending.count）；
metrics.UpdateQueueBytes（挂起的.sizeInBytes）；
//如果我们正在运行，并且捆绑包未满，请循环
//如果我们已关闭，请尝试发送待处理的内容
如果（！bundler.isReady（）&&！closed.get（））返回；
//表示我们将要发送已知大小的消息（以字节为单位）
metrics.incrementMessages（）；
incrementMessageBytes（bundler.sizeInBytes（））；

我知道这意味着bufferFull为false。但为什么每次都不是bufferFull，即使我经常调用URL

---

感谢您的关注。

请使用

Finchley.RELEASE

并设置

spring.zipkin.sender.type:kafka

谢谢您的建议。但是没有更改。您的yaml看起来格式不正确

 void flush(BufferNextMessage<S> bundler) {
  if (closed.get()) throw new IllegalStateException("closed");

  pending.drainTo(bundler, bundler.remainingNanos());

  // record after flushing reduces the amount of gauge events vs on doing this on report
  metrics.updateQueuedSpans(pending.count);
  metrics.updateQueuedBytes(pending.sizeInBytes);

  // loop around if we are running, and the bundle isn't full
  // if we are closed, try to send what's pending
  if (!bundler.isReady() && !closed.get()) return;

  // Signal that we are about to send a message of a known size in bytes
  metrics.incrementMessages();
  metrics.incrementMessageBytes(bundler.sizeInBytes());