Observability best practices for Lambda durable functions

This article covers observability best practices for AWS Lambda durable functions using a Stripe payment processing pipeline as an example.

• Monitor durable function-specific CloudWatch metrics like DurableExecutionDuration that include callback wait time
• Emit custom business metrics at each stage to create an end-to-end funnel tracking payment flow
• Define alarms for callback-specific failure modes: timeouts, webhook signature failures, and execution timeouts
• Use structured logging with correlation keys to trace payments across suspension boundaries and separate invocations
• Add X-Ray annotations at business-critical steps to enable searchable tracing of payment intents and callbacks
• Leverage the Lambda console Durable executions tab to view step-by-step timelines and callback states
• Distinguish between per-callback timeouts (external system issues) and whole-execution timeouts (configuration issues)

Layering custom metrics, structured logging, X-Ray tracing, and the durable executions tab provides complete visibility into long-running workflows that suspend on external callbacks.