Users report that some requests to a Lambda-backed feature fail with no code exception, and the function's own error count is flat. Which CloudWatch metric shows whether invocations are being rejected because of concurrency limits?
Show answer and explanation
Answer: C
The Throttles metric counts invocation requests that Lambda rejected because the function or account reached its concurrency limit. Those requests never ran, so the Errors metric does not show them. Invocations counts runs that Lambda actually performed, and IteratorAge tracks how far behind a stream consumer is, which is unrelated to concurrency rejection.
Why the other options are wrong
- A. Invocations counts the times the function was run, including successful and failed runs. Throttled requests were not run, so this metric will not point to the rejected calls.
- B. Errors counts invocations that ran and then failed, such as unhandled exceptions or timeouts. A throttled request never started, so it does not appear as an error.
- D. IteratorAge measures how old the newest record is when a stream consumer reads it. It describes stream processing lag rather than concurrency rejection of direct invocations.