ℹ️  SignalFx was acquired by Splunk in October 2019. See Splunk SignalFx for more information.

⚠️ Deprecation Notice

The SignalFx Tracing Library for Go is deprecated and will reach End of Support on June 8, 2023. After that date, this repository will be archived and no longer receive updates. Until then, only critical security fixes and bug fixes will be provided.

The SignalFx Tracing Library for Go helps you automatically instrument Go applications with instrumented library helpers and a tracer to capture and report distributed traces to SignalFx. You can also use the library to manually instrument Go applications with the OpenTracing API.

These are the requirements to use the SignalFx Tracing Library for Go:

• Go 1.12+
• Go Modules as dependency management

These are the supported libraries you can instrument:

Other libraries are available to instrument, but are in beta and aren't officially supported. You can view all the libraries in the contrib directory.

If the default configuration values don't apply for your environment, override them before running the process you instrument.

Follow these steps to instrument target libraries with provided instrumentors.

For more information about how to instrument a Go application, see the examples.

1. Import github.com/signalfx/signalfx-go-tracing to your go mod.

2. Import the instrumentor for the target library you want to instrument and replace utilities with their traced equivalents.

   Find an instrumentor for each supported target library in the contrib directory. Each instrumentor has an example_test.go file that demonstrates how to instrument a target library. Don't import examples directly in your application.

3. Enable tracing globally with tracing.Start(). This creates a tracer and registers it as the OpenTracing global tracer.

Link individual log entries with trace IDs and span IDs associated with corresponding events by passing metadata in HTTP reqeusts.

To inject trace context in HTTP requests, follow these steps:

1. Replace the default HTTP multiplexer:

   import (
     httptrace "github.com/signalfx/signalfx-go-tracing/contrib/net/http"
     "github.com/signalfx/signalfx-go-tracing/ddtrace/tracer" // global tracer
     "github.com/signalfx/signalfx-go-tracing/tracing" // helper
   )

2. Create the new HTTP multiplexer and attach it to your HTTP instance:

   mux := httptrace.NewServeMux()
   mux.HandleFunc("/", handler) // handler to your function
   // add all other routes
   http.ListenAndServe(":8888", mux)

3. Use a helper for the tracer to get the span ID and trace ID from each request:

   func TraceIdFromCtx(ctx context.Context) (result string) {
     if span, ok := tracer.SpanFromContext(ctx); ok {
       result = tracer.TraceIDHex(span.Context())
     }
     return
   }

   It cooperates with the global tracer instance.

This is an example of what injecting span IDs and trace IDs can look like in your environment:

package main

import (
  httptrace "github.com/signalfx/signalfx-go-tracing/contrib/net/http"
  "github.com/signalfx/signalfx-go-tracing/ddtrace/tracer" // global tracer
  "github.com/signalfx/signalfx-go-tracing/tracing" // helper

  "context"
  "fmt"
  "log"
  "net/http"
)

func TraceIdFromCtx(ctx context.Context) (result string) {
  if span, ok := tracer.SpanFromContext(ctx); ok {
    result = tracer.TraceIDHex(span.Context())
  }
  return
}

func handler(w http.ResponseWriter, r *http.Request) {
  ctx := r.Context()
  log.Print("Recevied request for ", r.URL.Path[1:], " ", TraceIdFromCtx(ctx))
}

func main() {
  tracing.Start()
  defer tracing.Stop()
  mux.HandleFunc("/", handler)
  http.ListenAndServe(":8888", mux)
}

To run integration tests locally, you should set the INTEGRATION environment variable. The dependencies of the integration tests are best run via Docker. To get an idea about the versions and the set-up take a look at our CI config.

The best way to run the entire test suite is using the CircleCI CLI. Simply run circleci build in the repository root. Note that you might have to increase the resources dedicated to Docker to around 4GB.