Closed
Description
I've seen numerous random CI failures like this one. Here's an another example.
The test failing is always test_vcs_url_urlquote_normalization
.
pip/tests/functional/test_install.py
Lines 485 to 502 in 285ff72
INFO pip._internal.vcs.bazaar:bazaar.py:40 Checking out http://bazaar.launchpad.net/%7Edjango-wikiapp/django-wikiapp/release-0.1 to /tmp/pytest-of-runner/pytest-1/popen-gw2/test_vcs_url_urlquote_normaliz0/cache/release-0.1
INFO pip.subprocessor:subprocess.py:119 Running command bzr checkout --lightweight --quiet http://bazaar.launchpad.net/%7Edjango-wikiapp/django-wikiapp/release-0.1 /tmp/pytest-of-runner/pytest-1/popen-gw2/test_vcs_url_urlquote_normaliz0/cache/release-0.1
INFO pip.subprocessor:subprocess.py:158 brz: ERROR: Connection error: while sending POST /~django-wikiapp/django-wikiapp/release-0.1/.bzr/smart: [Errno 110] Connection timed out
ERROR pip.subprocessor:subprocess.py:197 bzr checkout --lightweight --quiet http://bazaar.launchpad.net/%7Edjango-wikiapp/django-wikiapp/release-0.1 /tmp/pytest-of-runner/pytest-1/popen-gw2/test_vcs_url_urlquote_normaliz0/cache/release-0.1 exited with 3
A telltale sign is a singular job failing that takes >12 minutes longer than usual (as each of the four retries timeouts after ~270 seconds). This has been quite annoying. Is there any more reliable public Bazaar instance we can use? Actually, we can probably simply switch to using a public GitHub repository as the test doesn't need to use Bazaar AFAIK. The git blame goes all the way back to 2010.