🌱 e2e/apiexportendpointslice: use SharedKcpServer

[p0lyn0mial]

Summary

it is possible to use a shared server because the
scheduler skips shards annotated with experimental.core.kcp.io/unschedulable

Related issue(s)

Fixes #

[openshift-ci]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign jmprusi for approval. For more information see the Kubernetes Code Review Process.

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

Needs approval from an approver in each of these files:

• test/OWNERS

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

[fgiloux]

This looks good to me. I just have a couple of points:

Can you please amend the comment here?

kcp/test/e2e/reconciler/apiexportendpointslice/apiexportendpointslice_test.go

Lines 268 to 270 in 73cc8ef

	// Endpoint tests require the edition of shards.
	// These tests are run on a private cluster to avoid side effects on other e2e tests.
	// They require the resources previously created: APIExport, APIExportEndpointSlice, etc.

This could be a separate test function:

kcp/test/e2e/reconciler/apiexportendpointslice/apiexportendpointslice_test.go

Lines 342 to 382 in 73cc8ef

	t.Logf("Creating a slice without partition")
	sliceWithAll := &apisv1alpha1.APIExportEndpointSlice{
	ObjectMeta: metav1.ObjectMeta{
	GenerateName: "my-slice-without-partition",
	},
	Spec: apisv1alpha1.APIExportEndpointSliceSpec{
	APIExport: apisv1alpha1.ExportBindingReference{
	Path: exportClusterPath.String(),
	Name: slice.Spec.APIExport.Name,
	},
	},
	}
	sliceWithAll, err = sliceClient.Cluster(partitionClusterPath).Create(ctx, sliceWithAll, metav1.CreateOptions{})
	require.NoError(t, err, "error creating APIExportEndpointSlice")
	sliceWithAllName := sliceWithAll.Name
	t.Logf("Getting the total number of shards in the system")
	shards, err := shardClient.Cluster(core.RootCluster.Path()).List(ctx, metav1.ListOptions{})
	require.NoError(t, err, "error listing Shards")
	framework.Eventually(t, func() (bool, string) {
	sliceWithAll, err := kcpClusterClient.Cluster(partitionClusterPath).ApisV1alpha1().APIExportEndpointSlices().Get(ctx, sliceWithAllName, metav1.GetOptions{})
	require.NoError(t, err)
	if len(sliceWithAll.Status.APIExportEndpoints) == len(shards.Items) {
	for _, apiExportEndpoint := range sliceWithAll.Status.APIExportEndpoints {
	found := false
	for _, shard := range shards.Items {
	if strings.Contains(apiExportEndpoint.URL, shard.Spec.VirtualWorkspaceURL) {
	found = true
	break
	}
	}
	if !found {
	return false, fmt.Sprintf("didn't find matching shard for %v endpoint, shards: %#v", apiExportEndpoint.URL, shards.Items)
	}
	}
	return true, ""
	}
	return false, fmt.Sprintf("expected %d endpoint, but got: %#v", len(shards.Items), sliceWithAll.Status.APIExportEndpoints)
	}, wait.ForeverTestTimeout, 100*time.Millisecond, "expecting a single endpoint for the root shard, got %d", len(sliceWithAll.Status.APIExportEndpoints))

It was not the case previously due to the cost of setting up the private environment

[p0lyn0mial]

Updated the comment. IMO there is no need for extracting logic to a helper function. I like when tests are self-contained and self-descriptive.

[fgiloux]

Updated the comment. IMO there is no need for extracting logic to a helper function. I like when tests are self-contained and self-descriptive.

I did not mean a helper function. I meant a completely separate test function as it is not directly related to the tests part of the same function. It is not critical, just thinking it may be nicer.

[fgiloux]

I am concerned that you may be introducing a flake here: What does guarantee that no shard was created by a parallel test case between your call to list shards and the one to get the APIExportEndpointSlices?

[p0lyn0mial]

good point, assuming schedulable shards won't be removed we could change this logic to expect >= initial shards+ shards added by this tests.

[p0lyn0mial]

I did not mean a helper function. I meant a completely separate test function as it is not directly related to the tests part of the same function. It is not critical, just thinking it may be nicer.

ah, this could be a default (happy-path) scenario I always wanted, yeah I can do that :)

[mjudeikis]

@kcp-dev/kcp-contributors lets pick this up?

[kcp-ci-bot]

Issues go stale after 90d of inactivity.
After a furter 30 days, they will turn rotten.
Mark the issue as fresh with /remove-lifecycle stale.

If this issue is safe to close now please do so with /close.

/lifecycle stale