Python Support

info

In order to profile Python code you must have an interpreter with symbols.

You can check if your Python interpreter has symbols by running the following command:

$ nm /path/to/python | grep Py_BytesMain
❯  U Py_BytesMain

If you see Py_BytesMain then your Python interpreter has symbols and you can enable Python profiling.

tip

Sometimes the Python interpreter is dynamically linked as a shared library that has symbols, in that case you can check the shared library for symbols:

$ ldd /path/to/python
❯   linux-vdso.so.1 (0x00007fffccf70000)
    libpython3.11.so.1.0 => /path/to//libpython.so (0x00007f2bf8800000)
    libm.so.6 => /usr/lib/libm.so.6 (0x00007f2bf8713000)
    libc.so.6 => /usr/lib/libc.so.6 (0x00007f2bf8531000)
    /lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007f2bf8e68000)

$ nm /path/to/libpython.so | grep PyRuntime
❯   0000000000557d20 D _PyRuntime
    0000000000297290 T _PyRuntime_Finalize
    0000000000297220 T _PyRuntime_Initialize
    000000000029a7a0 T _PyRuntimeState_Fini
    000000000029a600 T _PyRuntimeState_Init
    000000000029a840 t _PyRuntimeState_ReInitThreads

If you have _PyRuntime then your Python interpreter has symbols and you can enable Python profiling.

Lastly, the official Python Docker -slim images do not have symbols because of space constraints, but you can copy the symbols from the "full" image into the "slim" image:

FROM python:3.11 as debuginfo
FROM python:3.11-slim
COPY --from=debuginfo /usr/local/lib/libpython3.11.so.1.0 /usr/local/lib/libpython3.11.so.1.0