This version is still in development and is not considered stable yet. For the latest stable version, please use Spring Boot 3.4.1! |
Tracing
Spring Boot Actuator provides dependency management and auto-configuration for Micrometer Tracing, a facade for popular tracer libraries.
To learn more about Micrometer Tracing capabilities, see its reference documentation. |
Supported Tracers
Spring Boot ships auto-configuration for the following tracers:
-
OpenTelemetry with Zipkin, Wavefront, or OTLP
-
OpenZipkin Brave with Zipkin or Wavefront
Getting Started
We need an example application that we can use to get started with tracing. For our purposes, the simple “Hello World!” web application that’s covered in the Developing Your First Spring Boot Application section will suffice. We’re going to use the OpenTelemetry tracer with Zipkin as trace backend.
To recap, our main application code looks like this:
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
@RestController
@SpringBootApplication
public class MyApplication {
private static final Log logger = LogFactory.getLog(MyApplication.class);
@RequestMapping("/")
String home() {
logger.info("home() has been called");
return "Hello World!";
}
public static void main(String[] args) {
SpringApplication.run(MyApplication.class, args);
}
}
There’s an added logger statement in the home() method, which will be important later.
|
Now we have to add the following dependencies:
-
org.springframework.boot:spring-boot-starter-actuator
-
io.micrometer:micrometer-tracing-bridge-otel
- bridges the Micrometer Observation API to OpenTelemetry. -
io.opentelemetry:opentelemetry-exporter-zipkin
- reports traces to Zipkin.
Add the following application properties:
-
Properties
-
YAML
management.tracing.sampling.probability=1
management:
tracing:
sampling:
probability: 1.0
By default, Spring Boot samples only 10% of requests to prevent overwhelming the trace backend. This property switches it to 100% so that every request is sent to the trace backend.
To collect and visualize the traces, we need a running trace backend. We use Zipkin as our trace backend here. The Zipkin Quickstart guide provides instructions how to start Zipkin locally.
After Zipkin is running, you can start your application.
If you open a web browser to localhost:8080
, you should see the following output:
Hello World!
Behind the scenes, an observation has been created for the HTTP request, which in turn gets bridged to OpenTelemetry, which reports a new trace to Zipkin.
Now open the Zipkin UI at localhost:9411
and press the "Run Query" button to list all collected traces.
You should see one trace.
Press the "Show" button to see the details of that trace.
Logging Correlation IDs
Correlation IDs provide a helpful way to link lines in your log files to spans/traces. If you are using Micrometer Tracing, Spring Boot will include correlation IDs in your logs by default.
The default correlation ID is built from traceId
and spanId
MDC values.
For example, if Micrometer Tracing has added an MDC traceId
of 803B448A0489F84084905D3093480352
and an MDC spanId
of 3425F23BB2432450
the log output will include the correlation ID [803B448A0489F84084905D3093480352-3425F23BB2432450]
.
If you prefer to use a different format for your correlation ID, you can use the logging.pattern.correlation
property to define one.
For example, the following will provide a correlation ID for Logback in format previously used by Spring Cloud Sleuth:
-
Properties
-
YAML
logging.pattern.correlation=[${spring.application.name:},%X{traceId:-},%X{spanId:-}]
logging.include-application-name=false
logging:
pattern:
correlation: "[${spring.application.name:},%X{traceId:-},%X{spanId:-}] "
include-application-name: false
In the example above, logging.include-application-name is set to false to avoid the application name being duplicated in the log messages (logging.pattern.correlation already contains it).
It’s also worth mentioning that logging.pattern.correlation contains a trailing space so that it is separated from the logger name that comes right after it by default.
|
Correlation IDs rely on context propagation. Please read this documentation for more details. |
Propagating Traces
To automatically propagate traces over the network, use the auto-configured RestTemplateBuilder
, RestClient.Builder
or WebClient.Builder
to construct the client.
If you create the RestTemplate , the RestClient or the WebClient without using the auto-configured builders, automatic trace propagation won’t work!
|
Tracer Implementations
As Micrometer Tracer supports multiple tracer implementations, there are multiple dependency combinations possible with Spring Boot.
All tracer implementations need the org.springframework.boot:spring-boot-starter-actuator
dependency.
OpenTelemetry With Zipkin
Tracing with OpenTelemetry and reporting to Zipkin requires the following dependencies:
-
io.micrometer:micrometer-tracing-bridge-otel
- bridges the Micrometer Observation API to OpenTelemetry. -
io.opentelemetry:opentelemetry-exporter-zipkin
- reports traces to Zipkin.
Use the management.zipkin.tracing.*
configuration properties to configure reporting to Zipkin.
OpenTelemetry With Wavefront
Tracing with OpenTelemetry and reporting to Wavefront requires the following dependencies:
-
io.micrometer:micrometer-tracing-bridge-otel
- bridges the Micrometer Observation API to OpenTelemetry. -
io.micrometer:micrometer-tracing-reporter-wavefront
- reports traces to Wavefront.
Use the management.wavefront.*
configuration properties to configure reporting to Wavefront.
OpenTelemetry With OTLP
Tracing with OpenTelemetry and reporting using OTLP requires the following dependencies:
-
io.micrometer:micrometer-tracing-bridge-otel
- bridges the Micrometer Observation API to OpenTelemetry. -
io.opentelemetry:opentelemetry-exporter-otlp
- reports traces to a collector that can accept OTLP.
Use the management.otlp.tracing.*
configuration properties to configure reporting using OTLP.
OpenZipkin Brave With Zipkin
Tracing with OpenZipkin Brave and reporting to Zipkin requires the following dependencies:
-
io.micrometer:micrometer-tracing-bridge-brave
- bridges the Micrometer Observation API to Brave. -
io.zipkin.reporter2:zipkin-reporter-brave
- reports traces to Zipkin.
Use the management.zipkin.tracing.*
configuration properties to configure reporting to Zipkin.
OpenZipkin Brave With Wavefront
Tracing with OpenZipkin Brave and reporting to Wavefront requires the following dependencies:
-
io.micrometer:micrometer-tracing-bridge-brave
- bridges the Micrometer Observation API to Brave. -
io.micrometer:micrometer-tracing-reporter-wavefront
- reports traces to Wavefront.
Use the management.wavefront.*
configuration properties to configure reporting to Wavefront.
Integration with Micrometer Observation
A TracingAwareMeterObservationHandler
is automatically registered on the ObservationRegistry
, which creates spans for every completed observation.
Creating Custom Spans
You can create your own spans by starting an observation.
For this, inject ObservationRegistry
into your component:
import io.micrometer.observation.Observation;
import io.micrometer.observation.ObservationRegistry;
import org.springframework.stereotype.Component;
@Component
class CustomObservation {
private final ObservationRegistry observationRegistry;
CustomObservation(ObservationRegistry observationRegistry) {
this.observationRegistry = observationRegistry;
}
void someOperation() {
Observation observation = Observation.createNotStarted("some-operation", this.observationRegistry);
observation.lowCardinalityKeyValue("some-tag", "some-value");
observation.observe(() -> {
// Business logic ...
});
}
}
This will create an observation named "some-operation" with the tag "some-tag=some-value".
If you want to create a span without creating a metric, you need to use the lower-level Tracer API from Micrometer.
|
Baggage
You can create baggage with the Tracer
API:
import io.micrometer.tracing.BaggageInScope;
import io.micrometer.tracing.Tracer;
import org.springframework.stereotype.Component;
@Component
class CreatingBaggage {
private final Tracer tracer;
CreatingBaggage(Tracer tracer) {
this.tracer = tracer;
}
void doSomething() {
try (BaggageInScope scope = this.tracer.createBaggageInScope("baggage1", "value1")) {
// Business logic
}
}
}
This example creates baggage named baggage1
with the value value1
.
The baggage is automatically propagated over the network if you’re using W3C propagation.
If you’re using B3 propagation, baggage is not automatically propagated.
To manually propagate baggage over the network, use the management.tracing.baggage.remote-fields
configuration property (this works for W3C, too).
For the example above, setting this property to baggage1
results in an HTTP header baggage1: value1
.
If you want to propagate the baggage to the MDC, use the management.tracing.baggage.correlation.fields
configuration property.
For the example above, setting this property to baggage1
results in an MDC entry named baggage1
.
Tests
Tracing components which are reporting data are not auto-configured when using @SpringBootTest
.
See Using Tracing for more details.