Common Timeout Propagation Mistakes
What you'll learn: The typical ways teams misconfigure or ignore timeout propagation, leading to wasted resources, cascading failures, and unpredictable system behavior.
The Problem
Even teams that understand timeout propagation often make critical mistakes in implementation. These pitfalls can turn your carefully designed timeout strategy into a source of failures rather than protection.
Classic Mistakes
1. Fixed Timeouts for All Hops
Using the same timeout value (e.g., 30 seconds) at every service in the chain. If Service A calls B calls C, and each has a 30-second timeout, the first hop could wait 30 seconds only to find the downstream already timed out at 29 seconds. You've wasted nearly a minute when the failure was determined in 30 seconds.
Analogy: Like giving every runner in a relay race the full marathon time limit, instead of accounting for legs already run.
2. Ignoring Propagation Entirely
Not passing timeout context at all. Each service starts with its own fresh timeout budget, causing services deep in the call chain to attempt work that the original caller has already abandoned.
3. Not Reserving Time for Retries
Setting remaining timeout exactly equal to downstream timeout. If you have 5 seconds left and immediately call downstream with a 5-second timeout, there's zero time for retry attempts, error handling, or response processing.
4. Forgetting Parallel Calls
When making parallel downstream calls, using the full remaining timeout for each call instead of the shared budget. If you have 10 seconds and fan out to 5 services each with 10-second timeouts, you're effectively claiming 50 seconds of budget.
5. Ignoring Clock Skew
Assuming all service clocks are perfectly synchronized when calculating remaining time, leading to negative timeout values or premature cancellations.
Key Takeaway: Timeout propagation fails when you use fixed values everywhere, ignore propagation entirely, don't budget for retries/processing, mishandle parallel calls, or ignore clock realities—each mistake wastes resources and undermines reliability.