Logs should help explain what happened, where it happened, and what action is needed. Unstructured messages and excessive sensitive detail make incidents harder to investigate and increase privacy risk.
Prefer structured events
Use consistent fields such as timestamp, severity, service, environment, request ID, event name, and outcome. Keep messages concise and use stable event identifiers for dashboards and alerts.
Add useful context carefully
Include identifiers that help correlate work, but avoid passwords, access tokens, session cookies, payment details, and raw sensitive request bodies. Hashing or truncating data is not automatically safe; decide whether the field is necessary at all.
Choose levels consistently
Use debug for temporary diagnostic detail, info for normal milestones, warning for recoverable unexpected conditions, and error for failures requiring investigation. Avoid logging the same exception at every layer.
Define retention and access
Restrict log access, set retention according to operational needs, and protect log transport and storage. Treat log exports as sensitive data.
Checklist
- Emit structured, consistent events.
- Include correlation identifiers.
- Redact secrets before logging.
- Use severity levels consistently.
- Restrict access and retention.