429 Too Many Requests vs 503 Service Unavailable

429 is caller-specific rate limiting. 503 is broader service unavailability from overload or maintenance.

Aspect	429 Too Many Requests	503 Service Unavailable
When it happens	Emitted when conditions specific to 429 too many requests occur.	Emitted when conditions specific to 503 service unavailable occur.
How to tell apart	Identify which layer emitted the code and inspect headers/logs for that layer.	Confirm whether problem is auth/policy, upstream quality, overload, or timeout.
Troubleshooting first step	Trace request ID through edge and application logs.	Check deploy/config changes and dependency health in same time window.
Practical note	Do not mask this response with generic handlers.	Return explicit headers like Retry-After or auth challenge when relevant.

When each appears in production

Use CDN/proxy logs to find emission point, then app logs/traces to isolate root cause.

Validate routing, policy, and timeout budgets in order. Fix the first failing layer before tuning client retries.