Client Library Features

Version 1.49 Latest

Jaeger clients have been retired. Please refer to this notice.

The table below provides a feature matrix for the existing client libraries. Cells marked with ? indicate that it’s not known if the given client supports the given feature and additional research & documentation update is required.

The table is auto-generated from data/clients.yaml

Data format and transport for reporting spans to Jaeger backend
Feature	Go	Java	Node.js	Python	C++	C#
Report `jaeger.thrift`, Compact protocol, over UDP	✅	✅	❌	✅	✅	✅
Report `jaeger.thrift`, Binary protocol, over UDP	❌	❌	✅	❌	❌	❌
Report `jaeger.thrift`, Binary protocol, over TCP (HTTP)	✅	✅	✅	❌	✅	✅
Report Zipkin Thrift, Binary protocol, over TCP (HTTP)	✅	✅	❌	❌	❌	❌
Inter-process propagation wire format (headers)
Feature	Go	Java	Node.js	Python	C++	C#
Uber’s original headers	✅	✅	✅	✅	✅	✅
Zipkin’s B3 headers	✅	✅	✅	✅	❌	✅
W3C Trace Context headers	coming	✅	coming	coming	coming	coming
Support inbound `jaeger-debug-id` header	when there is no trace	when there is no trace	❔	✅	❔	when there is no trace
Accept baggage from `jaeger-baggage` headers	when there is no trace	❌	❔	✅	❔	❌
Support for 128bit Trace ID	✅	✅	✅	✅	✅	✅
Metrics
Feature	Go	Java	Node.js	Python	C++	C#
Support standard tracer metrics (number of spans started, finished, reported, etc.)	✅	✅	✅	✅	❔	✅
Support standard RPC metrics	✅	❌	❌	❌	❌	❌
Metrics in Prometheus format	✅	✅	❌	✅	❔	❔
Tracer configuration
Feature	Go	Java	Node.js	Python	C++	C#
Support declarative tracer configuration	✅	✅	✅	✅	❔	✅
Allow configuring tracer tags, aka process tags	✅	✅	✅	✅	❔	✅
Allow remote configuration of samplers	✅	✅	✅	✅	✅	✅
Remotely configurable per-endpoint sampler	✅	✅	✅	✅	✅	✅
Remotely configurable baggage restrictions	✅	coming	coming	coming	❌	coming
Tracer configuration via environment variables
Feature	Go	Java	Node.js	Python	C++	C#
`JAEGER_SERVICE_NAME` defines service name that will be associated with the emitted spans.	✅	✅	✅	❌	✅	✅
`JAEGER_TAGS` defines a comma-separated list of static tags, aka “tracer tags”, e.g. `hostname=foobar,my.app.version=1.2.3`. These tags are added as the `Process` tags to each span.	✅	✅	✅	✅	✅	✅
`JAEGER_DISABLED` (not recommended) when set to true, instructs the Configuration to return a no-op tracer. Neither trace/span IDs nor baggage will be propagated with a no-op tracer. Instead of disabling the tracer completely, use `const` sampler with `0` param, which will minimize the overhead but keep the propagation going.	✅	❌	✅	❌	✅	❌
`JAEGER_AGENT_HOST` defines hostname for reporting spans over UDP/Thrift. To avoid packet loss, the agent is expected to run on the same machine as the application. This var is useful when there are multiple networking namespaces on the host.	✅	✅	✅	✅	✅	✅
`JAEGER_AGENT_PORT` defines port for reporting spans over UDP/Thrift.	✅	✅	✅	✅	✅	✅
`JAEGER_ENDPOINT` defines the URL for reporting spans via HTTP/Thrift. This setting allows for a deployment mode where spans are submitted directly to the collector.	✅	✅	✅	❌	✅	✅
`JAEGER_AUTH_TOKEN` defines an optional auth token when reporting over HTTP.	❌	✅	❌	❌	❌	✅
`JAEGER_USER` can be used for basic authentication when reporting over HTTP.	✅	✅	✅	❌	❌	✅
`JAEGER_PASSWORD` can be used for basic authentication when using reporting over HTTP.	✅	✅	✅	❌	❌	✅
`JAEGER_REPORTER_LOG_SPANS` instructs the Reporter to log finished span IDs. The reporter may need to be given a Logger for this option to take effect.	✅	✅	✅	❌	✅	✅
`JAEGER_REPORTER_MAX_QUEUE_SIZE` defines the max size of the in-memory buffer used to keep spans before they are sent out.	✅	✅	❌	❌	✅	✅
`JAEGER_REPORTER_FLUSH_INTERVAL` defines how frequently the report flushes span batches. Reporter can also flush the batch if the batch size reaches the maximum UDP packet size (~64Kb).	✅	✅	✅	❌	✅	✅
`JAEGER_SAMPLER_TYPE` defines the type of sampler to use, e.g. `probabilistic`, or `const` (see Sampling).	✅	✅	✅	❌	✅	✅
`JAEGER_SAMPLER_PARAM` provides configuration value to the sampler, e.g. `probability=0.001` (see Sampling).	✅	✅	✅	❌	✅	✅
`JAEGER_SAMPLER_MANAGER_HOST_PORT` defines the address of HTTP server that services client configuration, such as sampling strategies, baggage restrictions, throttling config, etc. The variable name is a legacy misnomer from the time when the server only provided the sampling strategies. At the moment only `jaeger-agent` implements this REST API.	✅	✅	✅	❌	✅	✅
`JAEGER_SAMPLER_REFRESH_INTERVAL` defines how often the sampler polls the config server for updates to the sampling strategies.	✅	❌	✅	❌	❌	❌
`JAEGER_SAMPLER_MAX_OPERATIONS` instructs the adaptive sampler to limit how many distinct operation names the sampler will track, to avoid unbound memory usage.	✅	❌	❌	❌	❌	❌
`JAEGER_PROPAGATION` defines the propagation format used by the tracer. Supported values are `jaeger` (defined here), `b3` (defined here) and `w3c` (defined here)	❌	✅	❌	❌	❌	✅
`JAEGER_TRACEID_128BIT` if true, instructs the tracer to generate 128bit trace IDs instead of the default 64bit. In the future 128bit will become the default.	✅	✅	❌	✅	❌	✅
`JAEGER_RPC_METRICS` when true, enables additional generation of RPC metrics from the tracing instrumentation. This is an experimental feature in the Go client. See also https://github.com/opentracing-contrib/java-metrics.	✅	❌	❌	❌	❌	❌