# How do I set up distributed tracing (OpenTelemetry) across microservices?

> Standardize on W3C Trace Context, leave propagation to the OTel SDK, and put traceparent in the queue message header, since that is where the chain snaps.

- Asked: 2026-05-06
- Answered: 2026-05-09
- Asked by: Ece
- Tags: mimari, observability, mikroservis
- Source: https://www.muhammetsafak.com.tr/en/just-ask/distributed-tracing-with-opentelemetry-across-laravel-and-go/
- Language: en-US
- Author: Muhammet Şafak

---
**Question:** A request travels API Gateway → Auth → Order → Stock. We're struggling to find the source of an 800ms delay in one of the services in between. With OpenTelemetry we plan to carry a `trace_id` across all services (Laravel and Go) and visualize it in a central APM (Jaeger/Grafana Tempo).

How do I design context propagation over HTTP headers and queue messages?


Short answer: don't invent a custom scheme; standardize on **W3C Trace Context** (`traceparent`/`tracestate`) so PHP and Go interoperate without friction.

Your real problem is "where's the 800ms" — and to see it the trace has to travel across services and queues **without breaking**. Here's the mechanism:

1. **Make W3C Trace Context the standard.** Start the root span at the gateway; every service **extracts** context from the incoming HTTP header and **injects** it into outgoing calls. Because it's a standard header, Laravel and Go read the same `traceparent` in common — don't invent your own `trace_id` header.
2. **Carry the same `traceparent` through queues.** When you publish a message, add `traceparent` as a message header/attribute and **re-extract** it in the worker. This is exactly the piece teams forget; skip it and async hops show up as **orphan traces** in the APM and the chain breaks.
3. **Put the OTel SDK in each service, ship via OTLP.** Service → OTLP → a **Collector** → Jaeger/Tempo. A Collector in between decouples export from the app and makes swapping backends easy.
4. **Choose sampling smartly.** Use `parent-based` so all spans of a trace share the same fate; add **tail sampling** on top so you definitely keep the 800ms outliers. With head-only sampling you drop the very slow trace you wanted to see.
5. **Don't hand-roll trace_id.** Don't write your own header read/write code; the SDK's context propagation already carries it. Do it by hand and you'll forget it somewhere and the trace snaps there.

**Bottom line:** I'd set up W3C Trace Context + the OTel SDK + a Collector + Tempo/Jaeger, leave propagation to the SDK, and never skip putting `traceparent` into the message header on the queue hop. I explain the architectural reasoning for why I build it this way on sade.dev; you'll find the detail of the Go-side HTTP mechanics in the hub post. Only an unbroken trace shows you the 800ms.

## Related Reading

- [Writing HTTP services in Go: is the standard library enough?](/en/blog/writing-http-services-in-go-is-the-standard-library-enough/) — Blog
- [When I Move to Microservices](https://sade.dev/en/journal/when-i-move-to-microservices) — sade.dev
- [Should I start a new project with microservices?](https://www.muhammetsafak.com.tr/en/just-ask/should-i-start-a-new-project-with-microservices/) — Just Ask
- [How do I peel the payment module off a PHP monolith with Strangler Fig?](https://www.muhammetsafak.com.tr/en/just-ask/peeling-the-payment-module-off-a-php-monolith-with-strangler-fig/) — Just Ask
- [For the message schema between Laravel and Go services, Protobuf or JSON Schema?](https://www.muhammetsafak.com.tr/en/just-ask/message-schema-between-laravel-and-go-protobuf-or-json-schema/) — Just Ask