You can now see how each Python function on AWS Lambda performs: every invocation with its duration, status and errors, which calls had a cold start, and traces that follow requests across functions and services. It is part of Atatus Serverless Monitoring.
You don't need to change your function code. Add the Atatus Python layer, set your license key, point the handler at the Atatus tracer and put your own handler name in ATATUS_ORIGINAL_HANDLER. It also works with the Serverless Framework.
What you can see
- Function performance: Track invocations, average and P75, P90 and P95 latency, and errors for each function over time.
- Error detection: Errors in your Python functions are captured with stack traces, so you can troubleshoot a failed invocation quickly.
- Cold starts: See which invocations had a cold start, so you can tell delays caused by startup from delays in your code.
- Invocation detail: Each invocation shows its timestamp, duration, trigger and status. Filter by duration to find the slowest calls.
- End-to-end tracing: Follow a request as it moves through several functions, queues and services in one trace.