Import from existing cluster

[196Ikuchil]

What type of PR is this?

/kind feature

What this PR does / why we need it:

The simulator will be able to import resources from an existing cluster indicated by a kubeconfig file.
kubeconfig is indicated by KUBECONFIG environment variable.
That import logic is enabled when environment variable EXTERNAL_IMPORT_ENABLED is "1".

【new feature】

• kubeconfig will be imported when the simulator is launched.
• ignoreRestart Option is implemented to ExportService for the Import function.
• Implement ReplicateExistingClusterService that has two ExportService.
  • The one is used for exporting the resources from the existing cluster.
  • Another one is used for importing those to the simulator.

Which issue(s) this PR fixes:

Fixes #41

Special notes for your reviewer:

The scheduler service of this simulator is not restarting when the Import function is executed.
This means that if we set kube scheduler config file of the existing cluster to the KUBE_SCHEDULER_CONFIG_PATH environment variable, we can apply the kube scheduler config file to the simulator. (we can't read the kube scheduler config from the existing cluster.)

About the import flow.

1. Do yarn start.
2. Read KUBE_SCHEDULER_CONFIG_PATH and KUBECONFIG environment variables and import the file.
3. Create many Services.
4. Start Scheduler. (this scheduler is used for a long time after this.)
5. Do Import function of this PR and import the existing cluster's resources to the scheduler.
6. Finish.

I tested this function for each resource on local and the import was successful.

• WIP:some test code.

/label tide/merge-method-squash
/hold

[k8s-ci-robot]

Hi @196Ikuchil. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

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.

[196Ikuchil]

This name is too long?

[sanposhiho]

np. 👍

[196Ikuchil]

Call the Import function of the ExportService to the simulator.
And it will not restart the simulator.

[196Ikuchil]

Is this good for this converting logic?
Actually, I want to convert a method like this.

kube-scheduler-simulator/scheduler/scheduler.go

Line 169 in 2802ed2

if err := scheme.Scheme.Convert(versioned, &cfg, nil); err != nil {

But I can't find it. Please give me some good ideas if you have any?

[lianghao208]

Yes, it's a good idea to use a generic conversion functions to convert internal object to applyConfiguration object.

Maybe you can refer to this extractor code:
https://github.com/kubernetes/kubernetes/blob/master/staging/src/k8s.io/pod-security-admission/admission/admission.go#L111-L134

In this case, it becomes:

func convertToApplyConfiguration(in interface{}) (interface{}, error) {
	_in, err := json.Marshal(in)
         if err != nil {
		return nil, xerrors.Errorf("call Marshal to cnvt object: %w", err)
	}
	switch in.(type) {
	case *corev1.Pod:
		p := &applyv1.PodApplyConfiguration{}
		if err := json.Unmarshal(_in, &p); err != nil {
			return nil, xerrors.Errorf("call Marshal to cnvt Pod: %w", err)
		}
		return p, nil
        case *corev1.Node:
                ...
        case *corev1.PersistentVolume:
                ...
        case *corev1.PersistentVolumeClaim:
                ...
	default:
		return nil, xerrors.Errorf("unknown type")
	}
	return nil
}

@sanposhiho WDYT?

[sanposhiho]

I'd suggest the following, which is similar to @lianghao208's idea, but pass the result to convertToApplyConfiguration as arg.

This eliminates the need for type conversions(from interface{} to XXXApplyConfiguration) in functions that call convertToApplyConfiguration.

func convertToApplyConfiguration(in interface{}, out interface{}) error {
	_in, err := json.Marshal(in)
	if err != nil {
		return xerrors.Errorf("call Marshal to cnvt object: %w", err)
	}
	switch in.(type) {
	case *corev1.Pod:
		typedout, ok := out.(*v1.PodApplyConfiguration)
		if !ok {
			return errors.New("unexpected type was given as out")
		}
		if err := json.Unmarshal(_in, &typedout); err != nil {
			return xerrors.Errorf("call Marshal to cnvt Pod: %w", err)
		}
		return nil
// other types as well .....
	default:
		return xerrors.Errorf("unknown type")
	}
	return nil
}

func cnvtPodListToApplyConfigurationList(pods []corev1.Pod) ([]v1.PodApplyConfiguration, error) {
	rto := make([]v1.PodApplyConfiguration, len(pods))
	for i, p := range pods {
		if err := convertToApplyConfiguration(p, &rto[i]); err != nil {
			return nil, xerrors.Errorf("convert pod to apply configuration: %w", err)
		}
	}
	return rto, nil
}

[196Ikuchil]

Thank you for both advice!
I am going to implement it based on @sanposhiho 's one.
it is also good for testing, I think.

I'll fix it tomorrow.

[196Ikuchil]

/assign @sanposhiho
/assign @lianghao208

[sanposhiho]

This time I only reviewed the comments and wording.
I'll review implementation details this weekend.

[sanposhiho]

/ok-to-test

[196Ikuchil]

Thanks for the review✋ Fixed it.

[sanposhiho]

Thanks. Just two nits!

[sanposhiho]

Now, this ConvertToResourcesForImportFromResourcesForExport is utility func. So, set expRes.SchedulerConfig here.
In replicateexistingcluster service, ResourcesForExport.SchedulerConfig from existing cluster should be already nil. So, just changing here as the above will work well.

[sanposhiho]

Please do this.

[sanposhiho]

Looks good overroll, left two more comments.

[sanposhiho]

remove this line

[sanposhiho]

Rename it to ConvertResourcesForImportToResourcesForExport

[sanposhiho]

/retest

[196Ikuchil]

Thanks!

[sanposhiho]

Move this logic to L62 (before starting simulator server)
If the server is started before the import is finished, you will see an empty resource list, depending on the timing of the front-end startup.

[sanposhiho]

@196Ikuchil Thanks for this huge work. Approved ❤️
(Also thanks @lianghao208 for reviewing this :) )

/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: 196Ikuchil, sanposhiho

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

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [sanposhiho]

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

[sanposhiho]

/unhold

[sanposhiho]

/retest