Kubevirt platform #779

nirarg · 2021-12-13T09:33:45Z

This PR adds KubeVirt platform as a new platform to hypershift cluster
This allows adding Kubevirt VMs as workers to hypershift cluster

This PR also introduce the new cluster-api-provider-kubevirt:
https://github.com/kubernetes-sigs/cluster-api-provider-kubevirt
The following fork is used for api and docker images creation: https://github.com/openshift/cluster-api-provider-kubevirt

netlify · 2021-12-13T09:34:29Z

✔️ Deploy Preview for hypershift-docs ready!

🔨 Explore the source changes: 0da4c20

🔍 Inspect the deploy log: https://app.netlify.com/sites/hypershift-docs/deploys/61c46558b817400008257103

😎 Browse the preview: https://deploy-preview-779--hypershift-docs.netlify.app/reference/api

enxebre · 2021-12-13T10:44:45Z

hypershift-operator/controllers/nodepool/nodepool_controller.go

 	return ctrl.Result{}, nil
 }

+func (r NodePoolReconciler) reconcileKubevirtMachineTemplate(ctx context.Context,


Let's put this into its own kubevirt.go file in this package.

Moved
Do you think have to keep this function under NodePoolReconciler struct?

enxebre · 2021-12-13T10:45:06Z

hypershift-operator/controllers/nodepool/manifests.go

 	}
 }

+func KubevirtMachineTemplate(infraName string, hostedCluster *hyperv1.HostedCluster, nodePool *hyperv1.NodePool, controlPlaneNamespace string) (*capikubevirt.KubevirtMachineTemplate, string) {


Let's put this into its own kubevirt.go file in this package.

enxebre · 2021-12-13T10:54:39Z

hypershift-operator/controllers/hostedcluster/internal/platform/platform.go

 		hcluster *hyperv1.HostedCluster,
 		controlPlaneNamespace string) error
+
+	AppendPlatformSpecificPolicyRules(rules []rbacv1.PolicyRule) []rbacv1.PolicyRule


Please, let's add a comment providing context on how this func should be implemented in the line of the ones for the other interface func.

Maybe we want to start with just allowing to return additional rules? In theory the current signature allows more dynamic introspection and addition, but not sure if it is needed, and people can just return nil instead of having to return what was passed in.

Changed it, please see 22e8c33

enxebre · 2021-12-13T11:04:50Z

hypershift-operator/controllers/hostedcluster/internal/platform/kubevirt/kubevirt.go

+
+func (p Kubevirt) CAPIProviderDeploymentSpec(hcluster *hyperv1.HostedCluster, tokenMinterImage string) (*appsv1.DeploymentSpec, error) {
+	defaultMode := int32(420)
+	capkLabels := map[string]string{


Let's drop this lables as we don't really need this labels, the generic func sets them.

enxebre · 2021-12-13T11:05:43Z

go.mod

 	sigs.k8s.io/cluster-api => sigs.k8s.io/cluster-api v1.0.0
 )
+
+replace sigs.k8s.io/cluster-api-provider-kubevirt => github.com/nirarg/cluster-api-provider-kubevirt-1 v0.0.0-20211212090220-ea93f714afd5


why is this coming from your personal fork?

As the capk repository is still in work in progress, not all changes are merged
Once all the changes from my local branch will get merged to openshift, will change it to use it

Changed to use the fork under openshift

enxebre · 2021-12-13T11:07:06Z

hypershift-operator/controllers/hostedcluster/internal/platform/kubevirt/kubevirt.go

+const (
+	hostedClusterAnnotation = "hypershift.openshift.io/cluster"
+	// TODO nargaman replace with pre-built valid image
+	imageCAPK = "registry:5000/capk:devel"


:face_with_raised_eyebrow:
Please let's update this to the proper image.

There is PR to add capk to openshift-ci, this would add the ability to generate the proper image
openshift/release#24575

Changed the image to one created from the fork in openshift org:
registry.ci.openshift.org/hypershift/cluster-api-kubevirt-controller:1.0.0

enxebre · 2021-12-13T11:08:33Z

Thanks @nirarg, this is looking great! dropped some comments mostly cosmetic.

rmohr

On a first look two minor comments.

rmohr · 2021-12-13T13:07:50Z

cmd/cluster/none/create.go

+	return nil
+}
+
+func SetCommon(exampleOptions *apifixtures.ExampleOptions, opts *core.CreateOptions) {


For my taste SetCommon does not look right. It is too unspecific and small to justify code-reuse and makes it harder to get which options are defaulted. What do you think?

rmohr · 2021-12-13T13:16:33Z

hypershift-operator/controllers/hostedcluster/internal/platform/platform.go

 		hcluster *hyperv1.HostedCluster,
 		controlPlaneNamespace string) error
+
+	AppendPlatformSpecificPolicyRules(rules []rbacv1.PolicyRule) []rbacv1.PolicyRule


Maybe we want to start with just allowing to return additional rules? In theory the current signature allows more dynamic introspection and addition, but not sure if it is needed, and people can just return nil instead of having to return what was passed in.

davidvossel

While we're waiting for some of the capk work to land, perhaps we can sort through how to automatically detect the rhcos containerdisk as well as the direction we want to go for making tunings like cpu/mem configurable on the VMIs

davidvossel · 2021-12-13T21:04:06Z

hosted-cluster-config-operator/controllers/resources/registry/registry.go

+	if platformType == hyperv1.NonePlatform || platformType == hyperv1.KubevirtPlatform {
 		cfg.Spec.Storage = imageregistryv1.ImageRegistryConfigStorage{EmptyDir: &imageregistryv1.ImageRegistryConfigStorageEmptyDir{}}
 	} else {
 		cfg.Spec.Storage.EmptyDir = nil


what happens if we don't set this type to EmptyDir and leave EmptyDir == nil for kubevirt platform?

davidvossel · 2021-12-13T21:21:06Z

hypershift-operator/controllers/nodepool/manifests.go

+						Domain: kubevirtv1.DomainSpec{
+							CPU:    &kubevirtv1.CPU{Cores: 2},
+							Memory: &kubevirtv1.Memory{Guest: &guestQuantity},
+							Devices: kubevirtv1.Devices{


What's the plan here for configuration? I'd imagine we need to expand the NodePoolPlatform struct in api/v1alpha1/nodepool_types.go with a KubeVirt platform type of some sort.

Is that something we should tackle in this PR? Maybe it could be as simple as exposing resource request/limits field so cpu+mem can be configured.

// NodePoolPlatform specifies the underlying infrastructure provider for the // NodePool and is used to configure platform specific behavior. type NodePoolPlatform struct { // Type specifies the platform name. // // +unionDiscriminator // +immutable Type PlatformType `json:"type"` // AWS specifies the configuration used when operating on AWS. // // +optional // +immutable AWS *AWSNodePoolPlatform `json:"aws,omitempty"` KubeVirt *KubeVirtNodePoolPlatform `json:"aws,omitempty"` } type KubeVirtNodePoolPlatform struct { // Requests is a description of the initial vmi resources. // Valid resource keys are "memory" and "cpu". // +optional Requests v1.ResourceList `json:"requests,omitempty"` }

after further discussion during our sync meeting today, we decided that we should embed the VMITemplate into the node pool for configuring the kubevirt platform. The rhcos volume/disk will be merged into the vmitemplate provided by the node pool

Changed it - added VMITemplate to the KubevirtPlatformNodePool

davidvossel · 2021-12-13T21:27:15Z

hypershift-operator/controllers/nodepool/manifests.go

+										// TODO nargaman - replace with pre-defined image
+										Image: "quay.io/containerdisks/rhcos:4.9",


With the aws provider, hypershift is attempting to automatically detect the correct AMI based on the release. If we are mirroring/publishing rhcos container disks to a predictable place that we control, maybe we follow the same pattern and search for images in the public quay.io repo.

Currently removed the container disk from the code and the user must provide this in CLI
When we will have more official image, will add it as default value

rmohr

@nirarg looks really great. A few minor suggestions.

rmohr · 2021-12-22T12:57:27Z

cmd/install/assets/hypershift_operator.go

+			Name: "credentials",
+			VolumeSource: corev1.VolumeSource{
+				Secret: &corev1.SecretVolumeSource{
+					SecretName: awsCredsSecretName,


Are there other AWS cases, like bare metal installations which make use of this secret too? If not, that change probably makes sense.

@enxebre can you please advise on this?

From what I can tell, PrivatePlatformType can only be none or aws right now. I think this should be fine the way you have it.

from cmd/install/install.go

func (o *Options) Validate() error { switch hyperv1.PlatformType(o.PrivatePlatform) { case hyperv1.AWSPlatform: if o.AWSPrivateCreds == "" || o.AWSPrivateRegion == "" { return fmt.Errorf("--aws-private-region and --aws-private-creds are required with --private-platform=%s", hyperv1.AWSPlatform) } case hyperv1.NonePlatform: default: return fmt.Errorf("--private-platform must be either %s or %s", hyperv1.AWSPlatform, hyperv1.NonePlatform) } return nil }

I have removed this change, as it not related to this PR

rmohr · 2021-12-22T13:00:49Z

api/fixtures/example.go

+			// Otherwise it must fail
+			// https://issues.redhat.com/plugins/servlet/mobile#issue/RFE-2501
+			if o.Kubevirt.Image == "" {
+				return nil, errors.New("The image for the Kubevirt machine must be provided by user (\"--kubevirt-image\" flag")


I think we expect that the commandline part already ensured that this value is populated. In fact I think that you are doing this alreay. You can probably remove this check and simplify the function return signature again.

Its can't be validated in the command line, because there is option to create the cluster without nodepool and create the nodepool later, in this case this parameter can be empty

But can't you then outside already verify that if the creation is requested, that you are also requiring this to be filled?

For example at https://github.com/openshift/hypershift/pull/779/files#diff-a8a4a70fd459a31d15b6500359d7e0e3cd47c5f7a48f551071756ddc395eaa20R55 ?

I just want to ensure that we keep the input validation together and don't spread it too much. The entrypoint above is supposed to see all persistent values filled already and should be able to handle your case based on the available info.

rmohr · 2021-12-22T13:02:22Z

cmd/cluster/kubevirt/create.go

+		Image:  "",
+	}
+
+	cmd.Flags().StringVar(&opts.KubevirtPlatform.Memory, "kubevirt-memory", opts.KubevirtPlatform.Memory, "In KubeVirt platform - the amount of memory which is visible inside the Guest OS (type BinarySI, e.g. 5Gi, 100Mi)")


How about node-memory or just memory? We are inside the kubevirt subcommand, so we do not in general need a kubevirt- prefix.

Changed to memory

rmohr · 2021-12-22T13:03:10Z

cmd/cluster/kubevirt/create.go

+	}
+
+	cmd.Flags().StringVar(&opts.KubevirtPlatform.Memory, "kubevirt-memory", opts.KubevirtPlatform.Memory, "In KubeVirt platform - the amount of memory which is visible inside the Guest OS (type BinarySI, e.g. 5Gi, 100Mi)")
+	cmd.Flags().Uint32Var(&opts.KubevirtPlatform.CPU, "kubevirt-cpu", opts.KubevirtPlatform.CPU, "In KubeVirt platform - Cores specifies the number of cores inside the vmi, Must be a value greater or equal 1")


How about node-cores or cores?

Changed it to Cores

rmohr · 2021-12-22T13:05:49Z

cmd/cluster/kubevirt/create.go

+
+	cmd.Flags().StringVar(&opts.KubevirtPlatform.Memory, "kubevirt-memory", opts.KubevirtPlatform.Memory, "In KubeVirt platform - the amount of memory which is visible inside the Guest OS (type BinarySI, e.g. 5Gi, 100Mi)")
+	cmd.Flags().Uint32Var(&opts.KubevirtPlatform.CPU, "kubevirt-cpu", opts.KubevirtPlatform.CPU, "In KubeVirt platform - Cores specifies the number of cores inside the vmi, Must be a value greater or equal 1")
+	cmd.Flags().StringVar(&opts.KubevirtPlatform.Image, "kubevirt-image", opts.KubevirtPlatform.Image, "In KubeVirt platform - Image is the name of the image with the embedded disk to be used to create the machines")


how about containerdisk, rootdisk, node-containerdisk with a descriptin like:

OpenShift compatible kubevirt containerdisk with ignition support (e.g. RHCOS, FCOS)

Changed it to containerdisk

rmohr · 2021-12-22T13:07:55Z

api/v1alpha1/nodepool_types.go

+// on KubeVirt platform.
+type KubevirtNodePoolPlatform struct {
+	// VirtualMachineInstance Spec contains the VirtualMachineInstance specification.
+	VirtualMachineInstanceSpec kubevirtv1.VirtualMachineInstanceSpec `json:"vmispec,omitempty"`


Hm, I don't think that we can use vmispec. How about template?

rmohr · 2021-12-22T13:12:13Z

cmd/nodepool/create.go

 	cmd.Flags().StringVar(&opts.RootVolumeType, "root-volume-type", opts.RootVolumeType, "The type of the root volume (e.g. gp2, io1) for machines in the NodePool")
 	cmd.Flags().Int64Var(&opts.RootVolumeIOPS, "root-volume-iops", opts.RootVolumeIOPS, "The iops of the root volume when specifying type:io1 for machines in the NodePool")
 	cmd.Flags().Int64Var(&opts.RootVolumeSize, "root-volume-size", opts.RootVolumeSize, "The size of the root volume (min: 8) for machines in the NodePool")
+	cmd.Flags().StringVar(&opts.KubevirtMemory, "kubevirt-memory", opts.KubevirtMemory, "In KubeVirt platform - the amount of memory which is visible inside the Guest OS (type BinarySI, e.g. 5Gi, 100Mi)")


here the kubevirt-* prefix is probably correct, but I think this is a part of the commandline tool which I did not address in the None platform refactoring. We probably want sub-commands for this too. I think this refactoring can be done independent of this PR @enxebre what do you think?

rmohr · 2021-12-22T13:13:13Z

go.mod

+	kubevirt.io/containerized-data-importer-api => github.com/kubevirt/containerized-data-importer-api v1.41.1-0.20211201033752-05520fb9f18d
 	sigs.k8s.io/apiserver-network-proxy/konnectivity-client => sigs.k8s.io/apiserver-network-proxy/konnectivity-client v0.0.24
 	sigs.k8s.io/cluster-api => sigs.k8s.io/cluster-api v1.0.0
+	sigs.k8s.io/cluster-api-provider-kubevirt => github.com/openshift/cluster-api-provider-kubevirt v0.0.0-20211222061349-10e498699601


The replaces look just how I have hoped that they will. 👍

rmohr · 2021-12-22T13:19:52Z

hypershift-operator/controllers/nodepool/nodepool_controller.go

+			return ctrl.Result{}, fmt.Errorf("failed to reconcile KubevirtMachineTemplate: %w", err)
+		}
+		span.AddEvent("reconciled kubevirtmachinetemplate", trace.WithAttributes(attribute.String("name", machineTemplate.GetName())))
+	case hyperv1.NonePlatform:


I think this is a rebase leftover. the None platform is handled in isAutomatedMachineManagement.

You right, removed it

rmohr · 2021-12-22T13:22:07Z

hypershift-operator/controllers/hostedcluster/internal/platform/kubevirt/kubevirt.go

+
+const (
+	hostedClusterAnnotation = "hypershift.openshift.io/cluster"
+	imageCAPK               = "registry.ci.openshift.org/hypershift/cluster-api-kubevirt-controller:1.0.0"


@nunnatsa 🥇

@nunnatsa can we change the tag? I think something like 0.0.1 or 0.0.1-prerelease would probably be more accurate since there is not even an alpha release of capik upstream.

sure. on it.

sure. on it.

I wonder if we can somehow delete the 1.0.0 tag from the openshift registry. It can potentially be a source of confusion for a long time :)

davidvossel · 2021-12-22T16:40:46Z

api/v1alpha1/nodepool_types.go

+// on KubeVirt platform.
+type KubevirtNodePoolPlatform struct {
+	// Template Spec contains the VirtualMachineInstance specification.
+	Template *kubevirtv1.VirtualMachineInstanceSpec `json:"template,omitempty"`


do you want to go ahead and adopt the full VMTemplate as I've outlined here? kubernetes-sigs/cluster-api-provider-kubevirt#61

If we're pretty confident this is the api we're going to move forward with. we could adopt it now on the hypershift side and then not have to deal with mutating the hypershift api in a backwards incompatible way later on.

the change on the hypershift api would look like this.

// VirtualMachineTemplateSpec defines the desired state of the kubevirt VM. type VirtualMachineTemplateSpec struct { // +kubebuilder:pruning:PreserveUnknownFields // +nullable ObjectMeta metav1.ObjectMeta `json:"metadata,omitempty"` // VirtualMachineSpec contains the VirtualMachine specification. Spec kubevirtv1.VirtualMachineSpec `json:"spec,omitempty" valid:"required"` } // KubevirtNodePoolPlatform specifies the configuration of a NodePool when operating // on KubeVirt platform. type KubevirtNodePoolPlatform struct { // Template Spec contains the VirtualMachine specification. VirtualMachineTemplate *VirtualMachineTemplateSpec `json:"virtualMachineTemplate,omitempty"` }

Until the capk PR merges, we'd simply only be using the virtualMachineTemplate.Spec.Template section when translating it to the capk KubeVirtMachine... but once capk adopts the full VMtemplate that would change.

I'm fine either way as long as making backwards incompatible changes to the hypershift NodePool api isn't going to cause us issues

Thanks @davidvossel
I changed it according to last capk change

rmohr

One nit.

rmohr · 2021-12-23T11:37:58Z

api/v1alpha1/nodepool_types.go

+// on KubeVirt platform.
+type KubevirtNodePoolPlatform struct {
+	// NodeTemplate Spec contains the VirtualMachineInstance specification.
+	NodeTemplate *capikubevirt.VirtualMachineTemplateSpec `json:"nodetemplate,omitempty"`


I think to be consistent the json filed should be named nodeTemplate (capital T).

rmohr · 2021-12-23T12:06:43Z

/lgtm

enxebre · 2022-01-03T12:27:47Z

hypershift-operator/controllers/nodepool/kubevirt_test.go

+	}
+}
+
+func generateNodeTemplae(memory string, cpu uint32, image string) *capikubevirt.VirtualMachineTemplateSpec {


nit: typo generateNodeTemplae

enxebre · 2022-01-03T12:29:11Z

cmd/nodepool/create.go

 		RootVolumeType: "gp2",
 		RootVolumeSize: 120,
 		RootVolumeIOPS: 0,
+		// TODO nargaman need to move this into platform specific create file


nit: usually we use this convention ordered by preference:
// TODO (jira-1234): desc
// TODO (username): desc
// TODO: desc

Changed my TODO comments to this format (// TODO (username): desc)

enxebre · 2022-01-03T12:29:27Z

cmd/cluster/kubevirt/create.go

+	}
+
+	cmd.Flags().StringVar(&opts.KubevirtPlatform.Memory, "memory", opts.KubevirtPlatform.Memory, "The amount of memory which is visible inside the Guest OS (type BinarySI, e.g. 5Gi, 100Mi)")
+	cmd.Flags().Uint32Var(&opts.KubevirtPlatform.Cores, "cores", opts.KubevirtPlatform.Cores, "The number of cores inside the vmi, Must be a value greater or equal 1")


"The number of cores inside the vmi, Must be a value greater or equal 1

Should we validate this below and fail otherwise? the way it's now zero would pass through right?

Added validation in applyPlatformSpecificsValues function

enxebre · 2022-01-03T12:30:57Z

hypershift-operator/controllers/hostedcluster/internal/platform/kubevirt/kubevirt_test.go

+				testNamespace,
+				apiendpoint)
+			if fnCallsCount != 1 {
+				t.Fatalf("Expected the provided fuction to be called once")


nit: typo fuction

enxebre · 2022-01-03T12:32:30Z

hypershift-operator/controllers/hostedcluster/hostedcluster_controller.go


-func reconcileCAPIProviderRole(role *rbacv1.Role) error {
-	role.Rules = []rbacv1.PolicyRule{
+func reconcileCAPIProviderRole(role *rbacv1.Role, p platform.Platform) error {


this is disrupting existing platform/behaviour, can we please have a unit test for it?

Added unit test, I used gomock to generate mock for the platform.Platform interface (gomock was also added to go.mod)

Did you run something manually to generate platform_generated.go that now we need to automate via Makefile?
If so may be just keep it simple for now, drop gomock and use kubevirt platform for the unit test.

This mock file is auto generated with go generate ./...
It can be added to the Makefile

ok, so we'd need to have a target for it and include a dep for the unit tests and verify targets. I'm not sure this is adding much value at this point since the interface and the logic are fairly simple. I'd rather just rely on kubevirt platform to unit test which adds also implicit value and avoid adding a new dep.

enxebre · 2022-01-03T12:34:43Z

hypershift-operator/controllers/hostedcluster/hostedcluster_controller.go


-func reconcileCAPIProviderRole(role *rbacv1.Role) error {
-	role.Rules = []rbacv1.PolicyRule{
+func reconcileCAPIProviderRole(role *rbacv1.Role, p platform.Platform) error {


consider p to be a pointer to avoid copying by value?

Changed the assigned structures to be pointers

I don't see the change, I'm still seeing you pass p platform.Platform) here. That's ok, this is not a blocker to me anyways.

enxebre · 2022-01-03T12:41:09Z

hypershift-operator/controllers/hostedcluster/hostedcluster_controller.go

 // components for a given platform.
 func (r *HostedClusterReconciler) reconcileCAPIProvider(ctx context.Context, createOrUpdate upsert.CreateOrUpdateFN, hcluster *hyperv1.HostedCluster,
-	capiProviderDeploymentSpec *appsv1.DeploymentSpec) error {
+	capiProviderDeploymentSpec *appsv1.DeploymentSpec, p platform.Platform) error {


consider p to be a pointer to avoid copying by value?

Changed the assigned structures to be pointers

I don't see the change, I'm still seeing you pass p platform.Platform) here. That's ok, this is not a blocker to me anyways.

enxebre · 2022-01-03T12:42:34Z

Thanks @nirarg, this looks great to me other than #779 (comment).
Also dropped a few nits.

enxebre · 2022-01-04T13:29:42Z

api/v1alpha1/nodepool_types.go

+// on KubeVirt platform.
+type KubevirtNodePoolPlatform struct {
+	// NodeTemplate Spec contains the VirtualMachineInstance specification.
+	NodeTemplate *capikubevirt.VirtualMachineTemplateSpec `json:"nodeTemplate,omitempty"`


I don't think we should omitempty here. Not a blocker.

enxebre · 2022-01-04T13:35:13Z

hypershift-operator/controllers/hostedcluster/internal/platform/platform.go

 	switch hcluster.Spec.Platform.Type {
 	case hyperv1.AWSPlatform:
-		platform = aws.AWS{}
+		platform = &aws.AWS{}


why this change?

This is done for this comment: #779 (comment)
The parameter passed in this function is interface, so changed the struct assignment to pointer

oh, sorry I was referring to the reconcileCAPIProvider signature where we could pass p as pointer instead, just a minor thought. I can't see how does this change makes any difference?

Yes, but p passed in the function is interface, you shouldn't change the interface to pointer but you can assign pointer to it, I think that do the job and prevent the copy, am I wrong?

…roller 1) Create capk-controller-manager 2) Add KubevirtCluster creation in hostedCluster reconcilation

* Add new KubevirtNodePoolPlatform in case platform type is KubeVirt * Create VirtualMachineTemplateSpec inside KubevirtNodePoolPlatform

enxebre · 2022-01-04T14:47:44Z

Thanks a lot @nirarg!
/lgtm
/approve

Let's follow up providing the nodepool cli platform awareness and getting into CI/docs.

openshift-ci · 2022-01-04T14:51:28Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: enxebre, nirarg

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~~ [enxebre]

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

openshift-ci · 2022-01-04T15:12:52Z

@nirarg: all tests passed!

Full PR test history. Your PR dashboard.

Details

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. I understand the commands that are listed here.

…en comment Minor fixes for code added in PR openshift#779 * Add kubevirt platform object to interface validations lines * Remove GoMock generate comment left in the code and not used

openshift-ci bot requested review from enxebre and sjenning December 13, 2021 09:34

enxebre reviewed Dec 13, 2021

View reviewed changes

rmohr reviewed Dec 13, 2021

View reviewed changes

davidvossel reviewed Dec 13, 2021

View reviewed changes

rmohr mentioned this pull request Dec 14, 2021

Discuss: Adding a KubeVirt Management Platform #591

Closed

6 tasks

openshift-ci bot added the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Dec 14, 2021

enxebre mentioned this pull request Dec 14, 2021

Introduce Agent platform #760

Merged

nirarg force-pushed the kubevirt-platform branch from 22e8c33 to 872ff11 Compare December 22, 2021 12:42

openshift-ci bot removed the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Dec 22, 2021

nirarg force-pushed the kubevirt-platform branch from 872ff11 to 8f35dba Compare December 22, 2021 13:08

rmohr suggested changes Dec 22, 2021

View reviewed changes

openshift-ci bot assigned rmohr Dec 22, 2021

nirarg force-pushed the kubevirt-platform branch 2 times, most recently from a9ae305 to b9c63b6 Compare December 22, 2021 16:11

davidvossel reviewed Dec 22, 2021

View reviewed changes

openshift-ci bot added the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Dec 22, 2021

nirarg force-pushed the kubevirt-platform branch from b9c63b6 to 1c74853 Compare December 23, 2021 11:30

openshift-ci bot removed the needs-rebase Indicates a PR cannot be merged because it has merge conflicts with HEAD. label Dec 23, 2021

rmohr reviewed Dec 23, 2021

View reviewed changes

nirarg force-pushed the kubevirt-platform branch from 1c74853 to 0da4c20 Compare December 23, 2021 12:02

openshift-ci bot added the lgtm Indicates that a PR is ready to be merged. label Dec 23, 2021

enxebre reviewed Jan 3, 2022

View reviewed changes

kubevirt-platform: Add cluster-api-provider-kubevirt assets

dd43a7c

nirarg force-pushed the kubevirt-platform branch from 59a6809 to c17ef0d Compare January 4, 2022 11:45

enxebre reviewed Jan 4, 2022

View reviewed changes

nirarg added 4 commits January 4, 2022 16:23

kubevirt-platform: Add "KubeVirt" platform type to HostedCluster cont…

6c7851f

…roller 1) Create capk-controller-manager 2) Add KubevirtCluster creation in hostedCluster reconcilation

kubevirt-platform: Add "KubeVirt" platform type to NodePool controller

ce62aa5

* Add new KubevirtNodePoolPlatform in case platform type is KubeVirt * Create VirtualMachineTemplateSpec inside KubevirtNodePoolPlatform

kubevirt-platform: Add "KubeVirt" platform type to the cmd

736f588

kubevirt-platform: Go modules changes

dea6ec3

nirarg force-pushed the kubevirt-platform branch from c17ef0d to dea6ec3 Compare January 4, 2022 14:24

openshift-ci bot assigned enxebre Jan 4, 2022

openshift-ci bot added the lgtm Indicates that a PR is ready to be merged. label Jan 4, 2022

openshift-ci bot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Jan 4, 2022

openshift-merge-robot merged commit 7d4fa7c into openshift:main Jan 4, 2022

nirarg mentioned this pull request Jan 10, 2022

Change NodePool CLI to have platform sub command #832

Merged

nirarg deleted the kubevirt-platform branch February 2, 2022 07:36

nirarg mentioned this pull request Feb 2, 2022

hostedcluster platform: Add kubevirt validation line and remove mockgen comment #952

Merged

4 tasks

		// TODO nargaman - replace with pre-defined image
		Image: "quay.io/containerdisks/rhcos:4.9",

Kubevirt platform #779

Kubevirt platform #779

Uh oh!

Conversation

nirarg commented Dec 13, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

netlify bot commented Dec 13, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

enxebre commented Dec 13, 2021

Uh oh!

rmohr left a comment

Choose a reason for hiding this comment

Uh oh!

rmohr Dec 13, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

davidvossel left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rmohr left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

nirarg commented Dec 13, 2021 •

edited

Loading

netlify bot commented Dec 13, 2021 •

edited

Loading

rmohr Dec 13, 2021 •

edited

Loading