問題文
After a failed deployment of the order API, an engineer needs to know which specific customer identifiers received an error response and what each error message said. The team already has counters for requests and errors. Which signal should the engineer reach for?
選択肢
- Metrics, because raising the scrape frequency to one second gives enough resolution to see each individual failed request separately.
- Traces, because a trace stores the full request and response bodies of every call, which includes the error message for each affected customer.
- Metrics, because the error counter can carry a label whose value is the customer identifier, letting the engineer read the identifiers straight off the graph.
- Logs, because each entry records one discrete event with its own message and attributes, which is what allows per-customer detail to be recovered.