A couple of times now, I've seen my buildbot consuming a constant 100%
CPU, despite not having an active build. It was a Python process
running inline code, and the code itself looked like it came from
Lib/test/test_profiling/test_sampling_profiler/test_integration.py
(specifically, I think it's the test_script as used by
test_subprocess, not the cli_test_script). A recent build seems to
have had a failure in that test:
https://buildbot.python.org/all/#/builders/345/builds/12988/steps/6/logs/wa…https://buildbot.python.org/all/#/builders/345/builds/12988/steps/6/logs/st…
which ended with a segfault.
Is this something that's likely to happen again in the future, and
thus should I devise a script that will find and kill rogue processes?
Or is this preventable within the test suite?
ChrisA