Spring-Boot-Reactive-RabbitMQ-Tracing-Example

Build

Required: Java 21, Maven, Docker

mvn clean install
docker-compose build

Run

docker-compose up -d
docker-compose logs -f downstream-service upstream-service

Test

[1. Partial, no DB-write] Call the Upstream service, which publishes on the queue for the Downstream service to consume and do nothing with.

curl http://localhost:12001/test/hello

[2. Partial, no Queue] Directly call the Downstream service, which saves the message in a DynamoDB.

curl http://localhost:12002/write/hello

[3. Full chain] Call the Upstream service, which publishes on the queue for the Downstream service to consume and save into a DynamoDB.

curl http://localhost:12001/write/hello

Upstream and Downstream service log using the same Trace-ID, which gets transmitted over RabbitMQ using the B3 or W3C header (depending on config from application properties), and all log entries have access to the Trace-ID.

URLs for debugging

Upstream service: localhost:12001

Downstream service: localhost:12002

RabbitMQ admin overview: http://localhost:15672/ (User: rabbitmq, Password: rabbitmq)

Localstack admin overview: https://app.localstack.cloud/

Details and explanation

Comments starting with !!! have been added in code with explanation of the critical steps.

Main learnings:

• Enable automatic context propagation wherever possible by adding Hooks.enableAutomaticContextPropagation() to the main method
• Enable observation for the RabbitMQ, which is disabled by default (see examples in code)
• Add .contextWrite(Function.identity()) just after creating a Mono from a non-reactive source
• Add .contextCapture() at the very end of the reactive pipeline in some obscure cases