[release-1.9] periodic-kubeadm-gce-1-9 building from release-1.8 branch

[enisoc]

There are two version tags. The main one seems to be pointing to the release-1.8 branch, which causes find_green_build to fail to find a good run.

Is this supposed to be a version-skewed test? The name doesn't indicate as such.

https://k8s-gubernator.appspot.com/build/kubernetes-jenkins/logs/periodic-kubernetes-e2e-kubeadm-gce-1-9/139

key	value
Result	SUCCESS
Tests	0 failed / 170 succeeded
Started	2017-11-30 11:20 PST
Elapsed	1h4m
Version	v1.8.5-beta.0.43+e22774fbe88837
Builder	6a55a3e2-d603-11e7-a235-0a580a6c001f
infra-commit	8b35d3f
job-version	v1.9.0-beta.0.713+5898890a735c4e
repo	k8s.io/kubernetes
repo-commit	e22774fbe88837403e507df24448bdb24c2b1f82
repos	{u'k8s.io/kubernetes': u'release-1.8'}
version	v1.9.0-beta.0.713+5898890a735c4e

[BenTheElder]

/cc @luxas @krzyzacy
I'll compare it to the configs for the 1.8 jobs.

[krzyzacy]

https://k8s-gubernator.appspot.com/build/kubernetes-jenkins/logs/periodic-kubernetes-bazel-build-1-9/143 - the periodic bazel build was failing now for release-1.9 :-(

[enisoc]

Filed kubernetes/release#473 for the bazel build.

[luxas]

Thanks @enisoc for filing, root cause seems to be #5576 (comment), basically a typo

@BenTheElder why doesn't periodic-kubernetes-e2e-kubeadm-gce-1-9 and periodic-kubernetes-e2e-kubeadm-gce-upgrade-1-8-1-9 seem to be present anymore in https://github.com/kubernetes/test-infra/blob/master/prow/config.yaml? Have they been deleted and if so for what reason? They were chained after the bazel 1.9 build job in the initial #5576 PR 🤔

[luxas]

@enisoc you're completely right, this job should solely test the creation of kubeadm clusters on the release-1.9 branch, no skew testing should be performed here.

[BenTheElder]

Yes, that job had the wrong branch and is fixed now thanks to @xiangpengzhao.
Unfortunately our config is large enough now that GitHub will no longer expand diffs which makes this fun to review :joy:

[BenTheElder]

@luxas I haven't removed any periodic jobs, the only thing I'm aware of is @krzyzacy changing the rate at which the bazel jobs run to reduce boskos pressure from the chained kubeadm jobs.