The test calls support.wait_process() which uses SHORT_TIMEOUT.

wait_process() should use LONG_TIMEOUT, or the ASAN buildbot should increase its timeout (regrtest --timeout parameter).

IMO using LONG_TIMEOUT is fine: it's ok if the test takes 2 minutes instead of 1 second, it's only important that it completes :-) The test should not measure the *performance* of the code, only if the code is valid. When tests are run in parallel, the buildbot system load can be very high. In this case, the system load was 1.70:

0:35:49 load avg: 1.70 [255/421/1] test_logging failed (1 failure) (1 min 18 sec)


AMD64 Arch Linux Asan Debug 3.10:
https://buildbot.python.org/all/#/builders/621/builds/466

======================================================================
FAIL: test_post_fork_child_no_deadlock (test.test_logging.HandlerTest)
Ensure child logging locks are not held; bpo-6721 & bpo-36533.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/buildbot/buildarea/3.10.pablogsal-arch-x86_64.asan_debug/build/Lib/test/test_logging.py", line 750, in test_post_fork_child_no_deadlock
    support.wait_process(pid, exitcode=0)
  File "/buildbot/buildarea/3.10.pablogsal-arch-x86_64.asan_debug/build/Lib/test/support/__init__.py", line 1971, in wait_process
    raise AssertionError(f"process {pid} is still running "
AssertionError: process 406366 is still running after 52.5 seconds