← Graph

Expected vs Unexpected Errors

concept 2 connections

Errors fall into two buckets: expected ones (known error bodies from a third party you've learned to handle) and unexpected ones (novel failures worth investigating). Reporting every expected error to tools like Rollbar or Sentry wastes developer cycles. The recommendation: only surface the unexpected ones to the team, and detect outages from expected errors via metrics and thresholds (e.g. percentage of failures) rather than per-exception alerts.

category
practice
about
Expected vs Unexpected Errors concept
Argues for splitting error reporting into expected (metrics/alarms) vs unexpected (team alerts).
about
Expected vs Unexpected Errors concept
Directly builds on the expected/unexpected distinction.

Provenance

Read by
4 extractions