Скачать или смотреть Manually Creating and Injecting TraceContext into Reactor Context in Spring Webflux

Learn how to effectively manage tracing in Spring Webflux by manually creating and injecting `TraceContext` to ensure consistent tracing across asynchronous calls.
---
This video is based on the question https://stackoverflow.com/q/76631278/ asked by the user 'Airidas36' ( https://stackoverflow.com/u/11535404/ ) and on the answer https://stackoverflow.com/a/76658896/ provided by the user 'Gaurav Bhatia' ( https://stackoverflow.com/u/8631387/ ) at 'Stack Overflow' website. Thanks to these great users and Stackexchange community for their contributions.

Visit these links for original content and any more details, such as alternate solutions, latest updates/developments on topic, comments, revision history etc. For example, the original title of the Question was: How to manually create and inject TraceContext into Reactor Context in Spring Webflux?

Also, Content (except music) licensed under CC BY-SA https://meta.stackexchange.com/help/l...
The original Question post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license, and the original Answer post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license.

If anything seems off to you, please feel free to write me at vlogize [AT] gmail [DOT] com.
---
Manually Creating and Injecting TraceContext into Reactor Context in Spring Webflux

In modern applications, especially those built on microservices architecture, tracing requests across various services is crucial for debugging and monitoring purposes. In this guide, we will explore the problem of reusing traceId from an initial REST request in a Spring Boot application that employs Webflux and RabbitMQ for asynchronous processing. We will also delve into potential solutions to ensure your tracing stays consistent throughout the application's lifecycle.

The Background

Picture this scenario: you have a Spring Boot 3.1.0 service that listens for REST requests and, upon receiving a request, publishes the payload to a RabbitMQ queue using reactor-rabbitmq. The response is returned almost instantly, but you also need to trace all subsequent API calls made to validate the data within this payload.

Initially, you can successfully capture the traceId and include it in the RabbitMQ message properties. However, the challenge arises when this traceId appears to be lost during later processing, and new traceIds are generated for downstream API calls. This inconsistency in tracing can lead to difficulties in tracking issues across your services.

The Problem

The issue at hand is that while you’re correctly setting the traceId in the RabbitMQ message properties, there seems to be a disconnect. After consuming the message, when you attempt to write this traceId into the Reactor's context, the WebClient ends up generating a new traceId for each call. This leads to confusion and makes it challenging to trace the flow of requests.

Is This Possible?

Yes, it is certainly feasible to preserve the traceId throughout your asynchronous operations; however, the implementation may require proper handling of the Spring Boot and Micrometer tracing configurations. Below are several potential solutions that can help address this challenge effectively.

Solution 1: Using Spring Cloud Binder for RabbitMQ

One of the simplest ways to maintain continuity in tracing across services is to utilize Spring Cloud Binder for RabbitMQ. It is designed to work with the reactive specifications natively, allowing your application to manage tracing seamlessly without much coding overhead.

Automatic Propagation: The integration will automatically read the trace/span information from request headers and include this in outgoing calls.

Context Management: The framework will handle context switching for you, taking care of most scenarios without extensive custom configurations.

Logging: If using MDC (Mapped Diagnostic Context), you can monitor trace information easily in logs.

Solution 2: Directly Write IDs into the Context

If integrating Spring Cloud Binder is not feasible for your project, consider directly writing specific identifiers (like traceId, spanId, etc.) into the Reactor context:

Use contextWrite(Context.of(TraceContext.class, yourTraceContext)) to add the relevant identifiers into the context.

Manually ensure that these identifiers are used when making downstream web calls.

Additionally, extracting valuable information from the context can be simplified by enabling:

[[See Video to Reveal this Text or Code Snippet]]

This minimal setup can help in ensuring the context flows naturally without manual threading.

Solution 3: Custom ReceiverContext Implementation

If the above solutions do not fit your needs, you can create a custom ReceiverContext that would manually handle observations:

Develop a new observation in your RabbitMQ message listener that incorporates all context info.

Leverage Micrometer's tools to extract headers automatically from