profiling

To identify high CPU usage, you can use performance profiling tools to sample the running process and generate a flame graph. This visualizes exactly where CPU time is spent, helping pinpoint the specific functions or code paths causing the spike.

Sampling periodically collects data on a program's state with low overhead, while instrumentation inserts code to measure exact execution times. Sampling is generally preferred for production environments to minimize impact, whereas instrumentation offers higher detail for specific functions.

Yes, async-profiler can be used to troubleshoot memory leaks in production systems by profiling memory allocation. It helps identify which objects are consuming the most memory and where they are being created in the code.

You can generate a flame graph by running a profiling tool like perf or py-spy against your application to collect stack traces. The profiling skill processes these traces to create a visual representation of time spent across code paths.

Yes, the profiling skill supports Ruby and Python applications through dedicated tools like rbspy and py-spy. These tools allow you to sample running processes and identify performance bottlenecks without requiring application modifications.