use failfast so we can see all the broken

[alex]

No description provided.

[alex]

Potential fix here: pypa/manylinux#1278

[alex]

Nope, that fix was not correct.

[mayeut]

@alex,
could not find anything related to a server outage for mirrors.fedoraproject.org which should be the origin for the error seen.
I see this happened twice 2 days ago, are you able to reproduce somewhere else (I can't). Might have been some transient network errors ?
Can you restart the failing workflows to see if this indeed disappeared ?
potential mitigation is probably to disable epel on your end if you don't need it. Something like that should work:

if yum-config-manager --version &> /dev/null; then yum-config-manager --disable epel; fi

[alex]

@mayeut Well, this is emberassing, looks like you're right that it was a flake that we over-interpreted.

[mayeut]

no worries.
FYI, at first I thought there was a typo in the PR you opened as we're seeing yum install/yum update occasional issues on manylinux2010 which already uses the vault... I was about to open pypa/manylinux#1281 when I saw your PR. I see that cryptography is planning dropping this as well pyca/cryptography#6694

[alex]

@mayeut yes pyca/cryptography#6831 documents our plan (the next release will be the last one to support manylinux2010 and the release after that will drop it)