WIP: More cluster-api

[justinsb]

Less hacky support for GCP, encode more of the logic into controllers.

WIP

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

Let's try this new test :-)

[hakman]

/test pull-kops-scenario-clusterapi-gcp

[hakman]

/test pull-kops-scenario-clusterapi-gcp

[hakman]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

This is an interesting one. For most builds, kops.Version is the same as kops.KOPS_RELEASE_VERSION, and we sometimes use KOPS_RELEASE_VERSION instead of kops.Version. But for CI builds, they are not the same. So when we use KOPS_RELEASE_VERSION with a CI build, we get an error like this one from kops-controller:

I1011 11:27:40.929602       1 server.go:247] bootstrap failed to build node config: building nodeConfig for instanceGroup: error during apply: kops version older than last used to update the cluster

*********************************************************************************

The cluster was last updated by kops version 1.34.0-beta.2+v1.34.0-beta.1-16-gce1ad2cc6c
To permit updating by the older version 1.34.0-beta.1, run with the --allow-kops-downgrade flag

*********************************************************************************

Trying to switch everything to use kops.Version

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

OK, the version thing changed image tags and had bigger impact. I'll try that separately, but I'll start with turning off the kops version check

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

Looks like we don't have the GCP project configured when we acquire it from boskos; fixing that!

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

Refactoring the signal handler, but something is odd there!

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

Working on a theory that killing go run doesn't send a signal to the child process (?), so trying to build it first.

The good news is that it seems to be working, so I think I could just move this code into toolbox or kops-controller, but ... let's see if this passes!

[justinsb]

/test pull-kops-scenario-clusterapi-gcp

I think this is getting pretty close :-)

[justinsb]

/retest

clusterapi test is now passing! Other failures look unrelated e.g. ( KUBE_BUILD_CONTAINER_NAME_BASE: unbound variable)

[k8s-ci-robot]

@justinsb: The following test failed, say /retest to rerun all failed tests or /retest-required to rerun all mandatory failed tests:

Test name	Commit	Details	Required	Rerun command
pull-kops-kubernetes-e2e-ubuntu-gce-build	059cb26	link	false	/test pull-kops-kubernetes-e2e-ubuntu-gce-build

Full PR test history. Your PR dashboard. Please help us cut down on flakes by linking to an open issue when you hit one in your PR.

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-sigs/prow repository. I understand the commands that are listed here.

[justinsb]

(I am still skipping the version check, which should not be needed, but I think this is #17658)