A production API with Axum and SQLx

What you end up with: a short-link API that survives contact with production — Postgres with migrations, request validation, structured errors, tracing with request IDs, integration tests against a real database, Docker, and a deploy.

This is the track that matches what Rust backend jobs actually ask for. Most tutorials stop at “here is a handler that returns JSON”. The parts that take real work — connection pooling, error mapping, migrations, testing against a live database, deploying — are exactly the parts covered here.

What you’ll build

• POST /links creates a short link, GET /:code redirects and counts the click
• GET /links lists links with cursor pagination
• Postgres schema managed by sqlx-cli migrations
• Errors as a single ApiError type, mapped to correct status codes with IntoResponse
• Input validation that returns useful messages instead of 500s
• tracing spans with request IDs, and JSON logs in production
• Integration tests that spin up a database per test
• A multi-stage Dockerfile and a deploy to a real host

Before you start

• Option, Result, and error handling — this track leans on custom error types
• Traits and generics — enough to read extractor bounds
• Concurrency and async — async/await and Send basics
• Comfortable with SQL, and with cargo from the command line

Articles

• What we’re building: a short link APIthe contract, then the fifteen steps to it
• Setting up Axum and your first routeaxum::serve and a /health that answers
• Routing and extracting path and query parametersextractors, and the order they must be in
• Returning JSON and building a response typeone envelope, one IntoResponse
• Adding SQLx and connecting to Postgresthe pool, and the wrong DATABASE_URL you will write
• Writing your first migrationsqlx migrate, and why the file is plain SQL
• Queries that the compiler checksquery! versus query, and sqlx prepare
• Inserting and reading linksthe two SQL paths the whole service is
• Generating short codes and handling collisionsbase62, a unique index, and a retry
• Validating input and returning proper errorsfrom a bad URL to the right status code
• API key authentication with middlewarea tower layer, and errors from inside it
• Rate limiting per API keya token bucket, and why one instance is not enough
• Structured logging and request tracingone request id through every line
• Integration tests against a real databasesqlx::test and a schema per test
• Docker, config, and deploying to productiona multi-stage build and a single binary

Where the code lives

axum-sqlx-api/ in the examples repository — one tag per article, plus an answers branch for the exercises.