log-aggregation

Log Aggregation

Collecting, indexing, and querying logs for debugging and monitoring.

Context

You are setting up log aggregation. Use structured logs; make them queryable.

Domain Context

• Structured: JSON logs with fields; queryable not text
• Trace ID: Correlate logs across services
• Centralized: Logs from all services in one place
• Retention: How long to keep logs? Cost/value tradeoff
• Searchability: Fast queries for debugging

Instructions

1. Structured Logging: JSON with fields not free text
2. Trace ID: Include on every log; correlate across services
3. Log Levels: DEBUG, INFO, WARN, ERROR; use appropriately
4. Centralization: ELK, Datadog, Splunk; pick one
5. Filtering: Useful queries to narrow down issues
6. Retention: Keep recent logs accessible; archive old
7. Performance: Don't log too much; expensive and slow

Anti-Patterns

• Text logs without structure; can't query programmatically
• No trace IDs; can't correlate across services
• Logging everything; too much data, hard to find signal
• Keeping logs forever; expensive storage
• No log pipeline; raw logs, no transformation/enrichment

Further Reading

• ELK Stack documentation
• Structured logging guide
• Log aggregation best practices (Google Cloud)