Track Errors First

a year ago

Observability traditionally focuses on dashboards, traces, metrics, and logs, but often overlooks errors, which are the most valuable signal.
Exceptions indicate a failure in the code's assumptions, making them the most useful event to track in observability.
The three pillars of observability (logs, metrics, traces) are useful but don't pinpoint where the code broke; proper error tracking is needed for that.
Exceptions provide high-signal events with direct information about failures, including stack traces, local variables, request data, and user context.
Error tracking is often abstracted in observability platforms, treating exceptions as just another data point rather than meaningful failures.
Many APM tools claim to track errors but often only count them, lacking the full context needed to understand and fix issues.
Some error tracking tools have evolved into full observability platforms, adding noise and reducing focus on errors.
The author, Klaas, founder of Bugsink, advocates for prioritizing error tracking in observability, arguing that current trends push errors to the background.

Hasty Briefsbeta