Detect --extra-config issues immediately rather than waiting 10 minutes

[tstromberg]

If a user feeds bad data into apiserver, the apiserver crashes immediately with messages that can be found using minikube logs --problems.

 ~  minikube logs --problems                                  Thu 18 Jul 2019 09:29:35 AM PDT
❌  Problems detected in "kube-apiserver":
    error: invalid audit log mode 777, allowed modes are "batch,blocking,blocking-strict"
❌  Problems detected in "kube-addon-manager":
    error: unable to recognize "STDIN": Get https://localhost:8443/api?timeout=32s: dial tcp 127.0.0.1:8443: connect: connection refused
    error: unable to recognize "STDIN": Get https://localhost:8443/api?timeout=32s: dial tcp 127.0.0.1:8443: connect: connection refused
    error: unable to recognize "STDIN": Get https://localhost:8443/api?timeout=32s: dial tcp 127.0.0.1:8443: connect: connection refused

This makes experimentation hard. Instead of telling users about it immediately, we currently make them wait ~8 minutes. We can do better :)

[tstromberg]

Example command-line:

minikube start --extra-config=apiserver.audit-dynamic-configuration=true --extra-config=apiserver.audit-policy-file=- --extra-config=apiserver.audit-log-path=- --extra-config=apiserver.audit-log-mode=777 --extra-config=apiserver.audit-log-maxage=5 --extra-config=apiserver.audit-log-maxbackup=5 --feature-gates=DynamicAuditing=true

[tstromberg]

See #4488 and any of these issues for inspiration:

https://github.com/kubernetes/minikube/issues?utf8=✓&q=label%3Acause%2Fapiserver-flags+

[olivierlemasle]

This delay before failure is actually due to Kubeadm. I logged an issue there, that was closed because it's not currently possible: kubernetes/kubeadm#1690

[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

[tstromberg]

Still an issue in minikube v1.6.

When we are waiting for the apiserver to become healthy, it would be nice to notice these failures and surface them in real time. Help wanted!