austin-tui 命令

常用示例

说明

austin-tui is a top-like text-based user interface for Austin, the frame stack sampler for CPython. It provides a real-time terminal view of where a Python application spends its time or allocates memory, displaying CPU and memory usage plots alongside per-frame statistics. Austin works by reading the CPython interpreter's virtual memory to sample frame stacks, requiring zero instrumentation and imposing minimal overhead on the target application. austin-tui wraps this sampling data in a curses-based interface inspired by the Unix top command. The default view shows the last-seen stack for the current process and thread, refreshing approximately once per second. Pressing F switches to full mode, which displays a hierarchical tree of all collected frame statistics. Pressing G activates a live flame graph visualization. Thread navigation is done with the left and right arrow keys, with the current PID and TID shown in the header. Additional interactive keys include P to pause the display while sampling continues in the background, S to save collected profiling data to a timestamped file, and +/- to adjust the minimum percentage threshold for displayed frames. When running in a VS Code terminal, source file paths in the output support Ctrl+Click (Cmd+Click on macOS) navigation. Saved data can be loaded in the Austin VS Code extension for further flame graph analysis.

参数

-p _pid_, --pid _pid_

Attach to an existing Python process by its process ID.

-C, --children

Also sample child Python processes. Navigate between them with the arrow keys.

-m, --memory

Profile memory usage instead of wall-clock time.

-c, --cpu

Sample on-CPU stacks only, ignoring idle time.

-i _microseconds_, --interval _microseconds_

Set the sampling interval in microseconds (default is 100).

-x _seconds_, --exposure _seconds_

Sample for the given number of seconds only, then stop.

-t _milliseconds_, --timeout _milliseconds_

Approximate startup wait time; increase on slow machines (default is 100 ms).