Basic example showing distributed tracing across Finatra apps

This is an example app where two Finatra (Scala) services collaborate on an http request. Notably, timing of these requests are recorded into Zipkin, a distributed tracing system. This allows you to see the how long the whole operation took, as well how much time was spent in each service.

Here's an example of what it looks like

Implementation Overview

Web requests are served by Finatra controllers, which trace requests by default.

These traces are sent out of process with Zipkin Finagle integration via Http.

This example intentionally avoids advanced topics like async and load balancing, eventhough Finatra supports them.

Running the example

This example has two services: frontend and backend. They both report trace data to zipkin. To setup the demo, you need to start Frontend, Backend and Zipkin.

Once the services are started, open http://localhost:8080/

• This will call the backend (http://localhost:9000/api) and show the result, which defaults to a formatted date.

Next, you can view traces that went through the backend via http://localhost:9411/?serviceName=backend

• This is a locally run zipkin service which keeps traces in memory

Starting the Services

In a separate tab or window, start each of finatra.FrontendMain and finatra.BackendMain:

$ sbt "run-main -Dzipkin.initialSampleRate=1.0 finatra.FrontendMain"
$ sbt "run-main finatra.BackendMain"

Next, run Zipkin, which stores and queries traces reported by the above services.

wget -O zipkin.jar 'https://search.maven.org/remote_content?g=io.zipkin.java&a=zipkin-server&v=LATEST&c=exec'
java -jar zipkin.jar

Troubleshooting

Finagle includes a flag that dumps traces into the error log (or stderr in our example setup). This can be helpful if you want to see which annotations are being sent to zipkin.

ex.

$ sbt "run-main -Dzipkin.initialSampleRate=1.0 -Dcom.twitter.finagle.tracing.debugTrace=true finatra.FrontendMain"