e2e: strict --provider checking

[pohly]

Is this a BUG REPORT or FEATURE REQUEST?:

/kind cleanup

What happened:

In PR #68483 we tightened the --provider parameter checking of the e2e.test binary such that it only accepted known providers. This turned out to be problematic (#70058) because people had started to depend on the previous behavior, which was to accept anything and merely warn about it.

For the sake of giving people time to adapt, PR #70141 reverted to the older behavior.

What you expected to happen:

In 1.14 we want to allow only known providers. While at it, the error handling for unknown providers should be enhanced (print proper error message instead of calling Failf, which panics).

/assign
/sig testing

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[pohly]

@neolit123 kubeadm was one of the tools which got broken by the tightening of the --provider parameter checking. Does kubeadm still rely on the acceptance of unknown provider parameters or can the current warning be turned into an error now?

/remove-lifecycle stale

[neolit123]

@pohly yes the kubernetes-anywhere deployer is still bound to provider=kubernetes-anywhere (unknown) in test infra jobs.
https://github.com/kubernetes/test-infra/blob/master/config/jobs/kubernetes/sig-cluster-lifecycle/kubeadm-x-on-y.yaml#L25

we went back and forth on this topic, but if you wish to replace this support for unknown deployers i can switch those jobs to skeleton.
LMK.

[neolit123]

given one release has passed with the deprecated message on the CLI, it's probably safe to remove this functionality now. :)

[pohly]

"Lubomir I. Ivanov" <notifications@github.com> writes:

we went back and forth on this topic, but if you wish to replace this support for unknown deployers i can switch those jobs to `skeleton`.

Yes, please do. Once that is complete, I'll turn the warning into an error.

[neolit123]

done:
kubernetes/test-infra#10913