zipkin-junit5

This contains ZipkinExtension, a JUnit5 extension to spin-up a Zipkin server during tests.

ZipkinExtension aims to emulate a http collector. For example, it presents the v1 and v2 POST apis Zipkin Api, and supports features like gzip compression.

Usage

For example, you can write micro-integration tests like so:

java

@RegisterExtension
public ZipkinExtension zipkin = new ZipkinExtension();

// Pretend we have a traced system under test
TracedService service = new TracedService(zipkin.httpUrl(), ReportingMode.FLUSH_EVERY);

@Test
public void skipsReportingWhenNotSampled() throws IOException {
  zipkin.storeSpans(asList(rootSpan));

  // send a request to the instrumented server, telling it not to sample.
  client.addHeader("X-B3-TraceId", rootSpan.traceId)
        .addHeader("X-B3-SpanId", rootSpan.id)
        .addHeader("X-B3-Sampled", 0).get(service.httpUrl());

  // check that zipkin didn't receive any new data in that trace
  assertThat(zipkin.getTraces()).containsOnly(asList(rootSpan));
}

You can also simulate failures.

For example, if you want to ensure your instrumentation doesn't retry on http 400.

java

@Test
public void doesntAttemptToRetryOn400() throws IOException {
  zipkin.enqueueFailure(sendErrorResponse(400, "Invalid Format"));

  reporter.record(span);
  reporter.flush();

  // check that we didn't retry on 400
  assertThat(zipkin.httpRequestCount()).isEqualTo(1);
}

Besides httpRequestCount(), there are two other counters that can help you assert instrumentation is doing what you think:

collectorMetrics() - How many spans or bytes were collected on the http transport.

These counters can validate aspects such if you are grouping spans by id before reporting them to the server.