v1.8.1 is the final release. No new pull requests are accepted.
We urge all users to migrate to OpenTelemetry. Please refer to the notice in the documentation for details.
- Intended to be used with Jaeger backend, but can also be configured to send traces to Zipkin.
- Implements OpenTracing Java API.
- Supports Java 1.8 and above
The group ID com.uber.jaeger
has been deprecated and moved to a different repository.
Please switch to io.jaegertracing
, as the old group ID will be maintained only for bug fixes.
Please see CONTRIBUTING.md.
Click through for more detailed docs on specific modules.
- jaeger-client: the module that instrumented applications should usually include
- jaeger-core: the core implementation of the OpenTracing API
- jaeger-thrift: set of components that send data to the backend
- jaeger-zipkin: compatibility layer for using the Jaeger Tracer to Zipkin-compatible backends
- jaeger-micrometer: a metrics provider, to report internal Jaeger Client metrics to third-party backends, such as Prometheus
- jaeger-tracerresolver: an OpenTracing
TracingResolver
for the Jaeger Tracer.
All artifacts are published to Maven Central. Snapshot artifacts are also published to Sonatype. Follow these instructions to add the snapshot repository to your build system.
Please use the latest version:
In the usual case, you just need to include the following dependency to your project:
<dependency>
<groupId>io.jaegertracing</groupId>
<artifactId>jaeger-client</artifactId>
<version>$jaegerVersion</version>
</dependency>
This will bring a concrete sender, such as jaeger-thrift
, as well as the jaeger-tracerresolver
and jaeger-core
.
The Jaeger Java Client uses org.apache.thrift:libthrift:0.11.0
. By default, declaring a dependency on the
jaeger-thrift
module will bring a non-shaded version of Thrift (and others). A shaded version of the dependency is
available with the classifier shadow
.
Please see jaeger-core/README.
When testing tracing instrumentation it is often useful to make sure that all spans are being captured, which is not the case in production configurations where heavy sampling is applied by default. The following configuration can be provided to affect which sampling is applied to the new traces:
sampler:
type: const # can either be const, probabilistic, or ratelimiting
param: 1 # can either be an integer, a double, or an integer
The valid values for type
are:
const
: configures a sampler that always makes the same decision for new traces depending on theparam
: always no forparam=0
, always yes otherwise.probabilistic
: configures a sampler that samples traces with probability equal toparam
(must be between0.0
and1.0
)ratelimiting
: configures a samlper that samples traces with a certain rate per second equal toparam
The OpenTracing API defines a sampling.priority
standard tag that
can be used to affect the sampling of a span and its children:
import io.opentracing.tag.Tags;
Tags.SAMPLING_PRIORITY.set(span, 1);
Jaeger Tracer also understands a special HTTP Header jaeger-debug-id
,
which can be set in the incoming request, e.g.
curl -H "jaeger-debug-id: some-correlation-id" http://myhost.com
When Jaeger sees this header in the request that otherwise has no tracing context, it ensures that the new trace started for this request will be sampled in the "debug" mode (meaning it should survive all downsampling that might happen in the collection pipeline), and the root span will have a tag as if this statement was executed:
span.setTag("jaeger-debug-id", "some-correlation-id")
This allows using Jaeger UI to find the trace by this tag.