I would have wished to see some active involvement on the upstream pip ticket and maybe even a PR to address the problem. IMHO, spending time on CI workarounds may not prove the best approach, especially for issues like this.

I would invite others to become more involved with upstream project that affect us, especially those working on older platforms (hint centos7 and python 3.6).

As pip maintainers mentioned, they time does not worth working around python 3.6 bugs, especially as it is like two months away for EOL. Still, they expressed willingness to review PRs from others!!!

IMHO add the extra environment variables to all systems and containers is likely to be 10x more time consuming than fixing the problem upstream.

On Tue, 27 Jul 2021 at 16:38, Marios Andreou <marios@redhat.com> wrote:

this issue should now be resolved please reach out in the usual way if
there are further issues here

thanks

On Tue, Jul 27, 2021 at 10:01 AM Marios Andreou <marios@redhat.com> wrote:
>
> Hello tripleo o/
>
> new gate blocker for centos7 hitting stable/train tripleo repos
> https://zuul.opendev.org/t/openstack/builds?job_name=tripleo-ci-centos-7-content-provider
>
> bug is at https://bugs.launchpad.net/tripleo/+bug/1938079 - hitting
> stable/train (we only run c7 jobs there).
>
> investigation/resolution ongoing
>
> please hold rechecks for centos7 jobs.
>
> thank you