Test-infra: branchprotector: recently added jobs are not updated

[nikhita]

I recently added prowjobs for the publishing-bot in https://github.com/kubernetes/test-infra/pull/9895 and cluster-api-provider-digitalocean in https://github.com/kubernetes/test-infra/pull/9954.

The PR for cluster-api-provider-digitalocean was merged yesterday but the PR for publishing-bot was merged a week ago. For both these repos, I have noticed that the prowjobs don't show up as having the required status context.

This could lead to a PR getting merged if it gets /lgtm and /approve before all the tests run.

/kind bug

[nikhita]

cc @sttts
for publishing bot

cc @xmudrii @alvaroaleman
for cluster-api-provider-digitalocean

cc @cblecker
since you mentioned on Slack that you saw something similar for https://github.com/kubernetes-sigs/yaml and https://github.com/kubernetes/klog.

cc @BenTheElder
because you are working on https://github.com/kubernetes/test-infra/issues/10015

[nikhita]

Hmm, this is weird. https://github.com/kubernetes-sigs/cluster-api-provider-digitalocean/pull/120 is not getting merged because the lint test keeps failing (this test won't pass until https://github.com/kubernetes/test-infra/pull/10113 is merged).

I also noticed that the lint test does not show as required and the https://prow.k8s.io/pr?query=is%3Apr+repo%3Akubernetes-sigs%2Fcluster-api-provider-digitalocean+author%3Axmudrii+head%3Adep-fix says Unknown Merge Requirements. Any ideas about how to make these jobs truly blocking?

[cblecker]

/assign @fejta
/cc @cjwagner

CC'ing Cole and assigning Erick as oncall. Can we get some eyes on this one this week? I'm not sure what broke that's preventing the branch protector from doing it's thing, especially if it's also impacting tide's merge requirements. Did the on-cluster config get corrupted somehow?

[cjwagner]

Hmm, I'm not sure what is wrong with the branch protector, I don't see any errors in the logs and it appears to be running regularly. The last time the branchprotector touched the digitalocean repo was on the 8th at 10:49am and the publishing-bot repo has never been touched.

The in cluster Prow config is exposed here and seems to be fine: https://prow.k8s.io/config

Hmm, this is weird. kubernetes-sigs/cluster-api-provider-digitalocean#120 is not getting merged because the lint test keeps failing (this test won't pass until #10113 is merged).

Why is that behavior unexpected? That is the intended behavior unless skip_unknown_contexts is specified in the tide context config: https://github.com/kubernetes/test-infra/blob/master/prow/cmd/tide/config.md#context-policy-options

[fejta]

Let me take a look...

[cblecker]

branch protection settings don't seem to be right for klog or yaml, still:

• https://github.com/kubernetes/klog/settings/branch_protection_rules/2611641
• https://github.com/kubernetes-sigs/yaml/settings/branch_protection_rules/2667334

[fejta]

We're not toggling whether to use all branches/all repos correctly

[cblecker]

Verified that the fix is working as expected.

Thanks so much, @fejta!

/close

[k8s-ci-robot]

@cblecker: Closing this issue.

In response to this:

Verified that the fix is working as expected.

Thanks so much, @fejta!

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.