Skip to content
Switch branches/tags
Go to file

Build Status Released Version


The opentracing-spring-jaeger-starter simply contains the code needed to provide a Jaeger implementation of the OpenTracing's io.opentracing.Tracer interface.

For a project to be able to actually instrument a Spring stack, one or more of the purpose built starters (like io.opentracing.contrib:opentracing-spring-web-starter or io.opentracing.contrib:opentracing-spring-cloud-starter)
would also have to be included in the POM.

The opentracing-spring-jaeger-web-starter starter is convenience starter that includes both opentracing-spring-jaeger-starter and opentracing-spring-web-starter This means that by including it, simple web Spring Boot microservices include all the necessary dependencies to instrument Web requests / responses and send traces to Jaeger.

The opentracing-spring-jaeger-cloud-starter starter is convenience starter that includes both opentracing-spring-jaeger-starter and opentracing-spring-cloud-starter This means that by including it, all parts of the Spring Cloud stack supported by Opentracing will be instrumented

Library versions

Versions 1.x.y of the library are meant to target Spring Boot 2.x while versions 0.x.y are meant to be used with Spring Boot 1.5





Either dependency will ensure that Spring Boot will auto configure a Jaeger implementation of OpenTracing's Tracer when the application starts.

If no settings are changed, spans will be reported to the UDP port 6831 of localhost. The simplest way to change this behavior is to set the following properties:

for the UDP sender, or use an HTTP sender by setting the following property:

opentracing.jaeger.http-sender.url = http://jaegerhost:portNumber/api/traces

Configuration options

All the available configuration options can be seen in JaegerConfigurationProperties. The prefix to be used for these properties is opentracing.jaeger. Furthermore, the service name is configured via the standard Spring Cloud property.

Beware to use the correct syntax for properties that are camel-case in JaegerConfigurationProperties.

  • For properties / yaml files use -. For example opentracing.jaeger.log-spans=true
  • For environment variables use _. For example OPENTRACING_JAEGER_LOG_SPANS


If no configuration options are changed and the user does not manually provide any of the beans that the auto-configuration process provides, the following defaults are used:

  • unknown-spring-boot Will be used as the service-name if no value has been specified to the property or opentracing.jaeger.service-name (which has the highest priority).
  • CompositeReporter is provided which contains the following delegates:
    • LoggingReporter for reporting spans to the console
    • RemoteReporter that contains a UdpSender that sends spans to localhost:6831
  • ConstSampler with the value of true. This means that every trace will be sampled
  • NoopMetricsFactory is used - effectively meaning that no metrics will be collected


Configuring senders is as simple as setting a couple necessary properties

HTTP Sender

opentracing.jaeger.http-sender.url = http://jaegerhost:portNumber/api/traces

It's possible to configure authentication on the HTTP sender by specifying an username and password:

opentracing.jaeger.http-sender.username = username opentracing.jaeger.http-sender.password = password

Or by specifying a bearer token:

opentracing.jaeger.http-sender.authtoken = token

Note that when an HTTP Sender is defined, the UDP sender is not used, even if it has been configured

UDP Sender opentracing.jaeger.udp-sender.port=portNumber

Common cases

Set service name

Set to the desired name

Log Spans

By default spans are logged to the console. This can be disabled by setting:

opentracing.jaeger.log-spans = false

Additional reporters

By defining a bean of type ReporterAppender, the code has the chance to add any Reporter without having to forgo what the auto-configuration provides


  • Const sampler

    opentracing.jaeger.const-sampler.decision = true | false

  • Probabilistic sampler

    opentracing.jaeger.probabilistic-sampler.sampling-rate = value

    Where value is between 0.0 (no sampling) and 1.0 (sampling of every request)

  • Rate-limiting sampler

    opentracing.jaeger.rate-limiting-sampler.max-traces-per-second = value

    Configures that traces are sampled with a certain constant rate. For example, when sampler.param=2.0 it will sample requests with the rate of 2 traces per second.

  • Remote sampler

    Remote sampler consults Jaeger agent for the appropriate sampling strategy to use in the current service. This allows controlling the sampling strategies in the services from a central configuration in Jaeger backend. It can be configured like so:

The samplers above are mutually exclusive.

A custom sampler could of course be provided by declaring a bean of type io.jaegertracing.samplers.Sampler

Propagate headers in B3 format (for compatibility with Zipkin collectors)

opentracing.jaeger.enable-b3-propagation = true

Propagate headers in W3C Trace Context format

opentracing.jaeger.enable-w3c-propagation = true

Advanced cases

Manual bean provisioning

Any of the following beans can be provided by the application (by adding configuring them as bean with @Bean for example) and will be used to by the Tracer instead of the auto-configured beans.

  • io.jaegertracing.samplers.Sampler
  • io.jaegertracing.metrics.MetricsFactory

io.jaegertracing.Tracer.Builder customization

If arbitrary customizations need to be performed on Tracer.Builder but you don't want to forgo the rest of the auto-configuration features, TracerBuilderCustomizer comes in handy. It allows the developer to invoke any method of Tracer.Builder (with the exception of build) before the auto-configuration code invokes the build method. Examples of this type of customization can be seen in the B3CodecTracerBuilderCustomizer and ExpandExceptionLogsTracerBuilderCustomizer classes.


Beware of the default sampler in production

In a high traffic environment, the default sampler that is configured is very unsafe since it samples every request. It is therefore highly recommended to explicitly configure on of the other options in a production environment


Maven checkstyle plugin is used to maintain consistent code style based on Google Style Guides

./mvnw clean install

Tips and tricks

Completely disable tracing

There are times when it might be desirable to completely disable tracing (for example in a testing environment). Due to the multiple (auto)configurations that come into play, this is not as simple as setting opentracing.jaeger.enabled to false.

When one of the starters of this project is included, then io.opentracing.contrib:opentracing-spring-tracer-configuration-starter is also included since it performs some necessary plumbing. However, when opentracing.jaeger.enabled is set to false, then the aforementioned dependency provides a default Tracer implementation that needs the JAEGER_SERVICE_NAME environment variable (see this).

One simple way around this would be to do the add the following Spring configuration:

@ConditionalOnProperty(value = "opentracing.jaeger.enabled", havingValue = "false", matchIfMissing = false)
public class MyTracerConfiguration {

    public io.opentracing.Tracer jaegerTracer() {
        return io.opentracing.noop.NoopTracerFactory.create();

In the code above we are activating a io.opentracing.Tracer iff opentracing.jaeger.enabled is set to false. This tracer is necessary to keep the various Spring configurations happy but has been configured to not sample any requests, therefore effectively disabling tracing.

Trace id not propagated via the Feign client

If you are using Feign, in some cases it might be necessary to explicitely expose the Feign client in the Spring configuration, in order to get the uber-trace-id propagated. This can be done easily by adding the following into one of your configuration classes:

public Client feignClient() {
    return new Client.Default(null, null);


Follow instructions in RELEASE


Apache 2.0 License.