Course contentsShow
← System Design
Lesson 1141 of 1,91929. Observability: LoggingFree lesson

Log Levels in Structured Logs

Using standardized severity levels (DEBUG, INFO, WARN, ERROR, FATAL) as queryable fields.

Log Levels in Structured Logs

What you'll learn: How to use standardized severity levels as queryable fields in structured logs for efficient filtering and alerting.

The Core Idea

In structured logging, log levels aren't just text prefixes—they're queryable fields with standardized values. Instead of embedding severity in a message like "[ERROR] Database connection failed", you add a level or severity field set to a standard value like ERROR. This transforms filtering from regex pattern matching into simple field queries.

The Five Standard Levels

Most systems use these severity levels (from least to most critical):

  • DEBUG: Detailed diagnostic information for troubleshooting during development
  • INFO: Normal operational events (user logged in, job completed)
  • WARN: Potentially problematic situations that don't stop execution
  • ERROR: Serious problems that require attention but the service continues
  • FATAL: Critical failures that force the application to terminate

Why This Matters

When your logs are structured with "level": "ERROR" as a field, you can:

  • Query all errors instantly: level=ERROR (no regex needed)
  • Set up alerts that trigger only on ERROR or FATAL
  • Count warnings by service: SELECT COUNT(*) WHERE level=WARN GROUP BY service
  • Filter out debug noise in production without parsing text

The Analogy

Think of log levels like priority lanes at airport security. Instead of reading everyone's boarding pass to find first-class passengers (text parsing), you have separate, labeled lanes (queryable fields) that instantly categorize travelers.

Key Takeaway: Structured log levels transform severity from embedded text into queryable metadata, enabling fast filtering, precise alerting, and efficient analysis across millions of log entries.