observability

Structured logging formats log output as JSON, making logs machine-readable and searchable. Use Python's logging module with a JSON formatter to emit key-value pairs instead of plain text, enabling easier analysis of system behavior across services.

CPU profiling identifies which functions consume the most processing time. Use cProfile for statistical profiling, py-spy for low-overhead sampling, or line_profiler for line-by-line analysis to pinpoint bottlenecks in your code.

Post-mortem debugging captures variable states and stack traces at the point of failure without stopping execution. Enable it alongside structured logging to inspect what went wrong, then use pdb or ipdb to step through code and trace root causes.

Memory profiling tracks heap allocations and identifies memory leaks. memory_profiler shows memory consumption line-by-line, while timing decorators and context timers help correlate memory spikes with specific operations across your application.

Collect runtime metrics and logs from multiple services using JSON-formatted output and centralized aggregation. Combine structured logging with profiling data across components to track performance and troubleshoot issues spanning service boundaries.

Interactive debugging with breakpoint(), pdb, and ipdb enables step-by-step execution and variable inspection during development and post-mortem analysis. They complement logging and profiling by letting you pause execution and examine state directly at failure points.