Skip to content

Conversation

@vstinner
Copy link
Member

@vstinnervstinner commented Jan 22, 2022

Skip the 3 slowest tests of the Address Sanitizer CI of GitHub
Actions:

  • test_tools
  • test_peg_generator
  • test_concurrent_futures

These tests take between 5 and 20 minutes on this CI which makes this
CI job the slowest. Making this CI job faster makes the whole Python
workflow faster. These tests are run on all others CIs.

Example of Address Sanitizer output:

10 slowest tests: - test_peg_generator: 17 min 33 sec - test_tools: 8 min 27 sec - test_concurrent_futures: 5 min 24 sec - test_zipfile: 2 min 41 sec - test_compileall: 2 min 21 sec - test_asyncio: 2 min 17 sec - test_gdb: 1 min 43 sec - test_weakref: 1 min 35 sec - test_pickle: 1 min 18 sec - test_subprocess: 1 min 12 sec 

Moreover, test_concurrent_futures also seems to be affected by
bpo-45200 bug: libasan dead lock in pthread_create().

https://bugs.python.org/issue45200

Skip the 3 slowest tests of the Address Sanitizer CI of GitHub Actions: * test_tools * test_peg_generator * test_concurrent_futures These tests take between 5 and 20 minutes on this CI which makes this CI job the slowest. Making this CI job faster makes the whole Python workflow faster. These tests are run on all others CIs. Example of Address Sanitizer output: 10 slowest tests: - test_peg_generator: 17 min 33 sec - test_tools: 8 min 27 sec - test_concurrent_futures: 5 min 24 sec - test_zipfile: 2 min 41 sec - test_compileall: 2 min 21 sec - test_asyncio: 2 min 17 sec - test_gdb: 1 min 43 sec - test_weakref: 1 min 35 sec - test_pickle: 1 min 18 sec - test_subprocess: 1 min 12 sec Moreover, test_concurrent_futures also seems to be affected by bpo-45200 bug: libasan dead lock in pthread_create().
@vstinner
Copy link
MemberAuthor

vstinner commented Jan 22, 2022

@pablogsal: Are you ok with that?

Moreover, test_concurrent_futures also seems to be affected by bpo-45200 bug: libasan dead lock in pthread_create().

See: https://bugs.python.org/issue45200#msg411273

@pablogsal
Copy link
Member

@pablogsal: Are you ok with that?

Moreover, test_concurrent_futures also seems to be affected by bpo-45200 bug: libasan dead lock in pthread_create().

See: https://bugs.python.org/issue45200#msg411273

Yeah, go ahead 🤘

@vstinner
Copy link
MemberAuthor

Tests now take less than 10 minutes, that's better!

Tests / Address sanitizer (pull_request) Successful in 13m — Address sanitizer

== Tests result: SUCCESS == 406 tests OK. 10 slowest tests: - test_asyncio: 1 min 57 sec - test_venv: 1 min 22 sec - test_weakref: 1 min 13 sec - test_subprocess: 1 min 6 sec - test_gdb: 1 min 6 sec - test_pickle: 1 min 1 sec - test_compileall: 55.4 sec - test_signal: 54.6 sec - test_io: 50.9 sec - test_regrtest: 44.9 sec 9 tests skipped: test_devpoll test_ioctl test_kqueue test_msilib test_startfile test_winconsoleio test_winreg test_winsound test_zipfile64 Total duration: 9 min 47 sec Tests result: SUCCESS 

@vstinnervstinner merged commit ce7d667 into python:mainJan 22, 2022
@vstinnervstinner deleted the gha_skip_concurrent_futures branch January 22, 2022 18:16
@vstinner
Copy link
MemberAuthor

Ah, Python 3.10 doesn't have this CI job, so there is no need to backport it.

Yeah, go ahead metal

Thanks for the review. I merged my PR.

Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants

@vstinner@pablogsal@the-knights-who-say-ni@bedevere-bot