OTEL Tracing on GCP Cloud Run (Python FastAPI) → Google Cloud Trace

This guide shows how to instrument a Python FastAPI service running on GCP Cloud Run with OpenTelemetry (OTEL) and export traces to Google Cloud Trace. It includes the exact folder layout, code, and deploy/delete commands.

Suggested Google Sites Structure

• Home → “OTEL on GCP Cloud Run (FastAPI)”

• Prerequisites → Tooling & IAM

• Project Layout → Folder structure

• Source Code → Dockerfile, requirements.txt, FastAPI app, OTEL setup

• Deploy → Build & deploy commands

• Verify → Logs & Trace Explorer

• Toggles & Env Vars → Feature flags

• Troubleshooting → Common errors & fixes

Page navigation link names (short):

• Home

• Prereqs

• Layout

• Code

• Deploy

• Verify

• Env Vars

• Troubleshoot

Overview

• Goal: Capture function-level latency and external HTTP call latency from FastAPI on Cloud Run and view them in Cloud Trace Explorer.

• Approach: Use OTEL SDK + auto-instrumentation for fastapi, requests, httpx; export to stdout (for logs) and Cloud Trace.

• Why OTEL: Vendor-neutral, easy to add other exporters later (e.g., Datadog via OTLP/Collector).

Prerequisites

• GCP project with Cloud Run and Cloud Trace enabled

• Docker + gcloud CLI authenticated to your project

• Artifact Registry repo created (example uses: us-central1-docker.pkg.dev/sugunan/fastapi-repo)

• Cloud Run service account with Cloud Trace Agent (default on many projects)