[upodroid]

No description provided.

[upodroid]

This is working correctly, we need the linked PR in k/release to be merged first.

[hakman]

This is working correctly, we need the linked PR in k/release to be merged first.

I am having some trouble understanding what's happening here. Was there a discussion that I missed somewhere?

[upodroid]

This is working correctly, we need the linked PR in k/release to be merged first.

I am having some trouble understanding what's happening here. Was there a discussion that I missed somewhere?

It's a continuation of #18662 but it also works on the build-kubernetes flag which your PR didn't modify.

Also, getting kops e2e jobs to use ephemeral S3 buckets instead of gs://k8s-staging-kops/pulls has been a long-standing to-do item. (Eliminating the GCP egress fee as well as blocking presubmits from writing to the staging bucket.)

[upodroid]

The k/release PR has been merged. We can merge this PR now.

[hakman]

This is working correctly, we need the linked PR in k/release to be merged first.

I am having some trouble understanding what's happening here. Was there a discussion that I missed somewhere?

It's a continuation of #18662 but it also works on the build-kubernetes flag which your PR didn't modify.

Also, getting kops e2e jobs to use ephemeral S3 buckets instead of gs://k8s-staging-kops/pulls has been a long-standing to-do item. (Eliminating the GCP egress fee as well as blocking presubmits from writing to the staging bucket.)

Please try to understand, I am not aware of your intentions.
The PR description is BLANK, and this is a non-trivial change to review.

[hakman]

Thanks, the S3 path works end to end in the calico and ec2-build logs. A few issues inline.

For the description: besides S3 staging, this fixes --build-kubernetes on AWS. Master rejects the default gs://k8s-staging-kops/pulls/... path, which is what fails pull-kubernetes-ec2-master-scale-correctness-2000 today. It also makes pull-kops-kubernetes-e2e-ubuntu-ec2-build actually run on AWS.

[upodroid]

I applied all the suggestions.

[hakman]

One more small change, please.

[hakman]

This only covers the builder failing. A defer catches all cases.

func (d *deployer) Build() (err error) {
	// kubetest2 does not call Down() when Build() fails, so delete the staging bucket on
	// every error, not only when the builder fails.
	defer func() {
		if err != nil {
			err = errors.Join(err, d.deleteStagingStore())
		}
	}()

	if err := d.init(); err != nil {
		return err
	}
	results, err := d.BuildOptions.Build()
	if err != nil {
		return err
	}

[kubernetes-prow]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: hakman

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [hakman]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[hakman]

/test pull-kops-e2e-kubernetes-azure
/test pull-kops-e2e-kubernetes-do-dns-none
/test pull-kops-aws-upgrade-dns-none
/test pull-kops-e2e-aws-ipv6-external-dns
/test pull-kops-ec2-master-scale-performance-100
/hold for tests to run

[hakman]

/test pull-kops-e2e-k8s-aws-amazonvpc
/test pull-kops-aws-upgrade-dns-none

[upodroid]

/hold cancel
/retest

[upodroid]

/hold

[upodroid]

/test pull-kops-e2e-kubernetes-azure
/test pull-kops-e2e-kubernetes-do-dns-none
/test pull-kops-aws-upgrade-dns-none
/test pull-kops-e2e-aws-ipv6-external-dns
/test pull-kops-ec2-master-scale-performance-100

[upodroid]

/hold cancel

All the tests are passing

[hakman]

/lgtm

[hakman]

Thanks!