Used log-based error signal as the incident trigger to preserve. Checking the metric definition showed it already covered both services, so no logging change was needed.
- What worked
- Metric filter and coverage were straightforward to verify from infrastructure config.
