Skip to main content

Overview

  • DynamoAI components write structured application logs to standard output. Any log collector that reads Kubernetes container logs can send them to the backend of your choice.
  • This document covers the core logging concepts and structure used by DynamoAI components, which are independent of your chosen logging backend.

Log Collection Architecture​

  • DynamoAI components write logs to standard output; the DynamoAI chart does not install a log collector
  • In SaaS deployments: Logs are automatically exported to AWS CloudWatch
  • In PaaS deployments: Use the collector you already run, such as Fluent Bit, Fluentd, or the OpenTelemetry Collector, to send logs to any backend (CloudWatch, Elasticsearch, Splunk, etc.). For DynamoGuard moderation decisions, see Send DynamoGuard Moderation Logs To A SIEM.

Structured Logging​

DynamoAI uses structured logging across its components. Instead of printing logs as plain strings, logs are emitted as JSON objects with consistent attributes. This enables:

  • Easy filtering and querying based on metadata attributes
  • Simplified log exploration and debugging
  • Backend-agnostic log analysis (works with any logging system that supports JSON)

Benefits​

  • Filter logs by specific attributes (e.g., request ID, component, test ID) rather than searching through log text
  • Consistent log structure across all DynamoAI components
  • Better integration with modern logging backends and analysis tools

Component Logging Documentation​