Course contentsShow
← System Design
Lesson 1013 of 1,91926. Idempotency and RetriesFree lesson

Handling In-Progress Requests

Detect when a duplicate request arrives while the original is still processing and return appropriate status codes or wait.

Handling In-Progress Requests

What you'll learn: How to detect and handle duplicate requests that arrive while the original request is still being processed.

The Problem

You've implemented idempotency keys to prevent duplicate operations, but what happens when a client retries while the first request is still running? If you simply check whether the idempotency key exists, you might miss that the operation hasn't completed yet. The client could receive inconsistent responses or trigger race conditions.

The Core Pattern

When a duplicate request arrives mid-processing, you have two main strategies:

1. Return a "Processing" Status

Store the idempotency key immediately when processing begins, marking it as "in-progress." When a duplicate arrives, check this status:

  • Status = in-progress: Return HTTP 409 Conflict or HTTP 202 Accepted with a message like "Request is being processed"
  • Status = completed: Return the cached result from the first request

2. Wait and Return the Actual Result

Some systems hold the duplicate request open until the original completes, then return the same result to both. This requires tracking pending requests and notifying waiting clients when processing finishes.

Storage Pattern

Your idempotency key record needs at least three states:

  • processing (request started but not finished)
  • completed (finished successfully, result cached)
  • failed (finished with error, may allow retry)

Analogy: Think of a restaurant kitchen with order tickets. If a waiter tries to submit the same order twice while the first is still cooking, the system should say "already cooking" rather than starting a duplicate meal.

Key Takeaway: Handling in-progress requests requires tracking request lifecycle states beyond just "done or not done," preventing race conditions and giving clients meaningful feedback during concurrent retries.