Skip to content

CNTRLPLANE-3553: Wire osImageStream into NodePool controller (hash, token, status, validation) #8730

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sdminonne wants to merge 2 commits into
base: main
Choose a base branch
from
Open

CNTRLPLANE-3553: Wire osImageStream into NodePool controller (hash, token, status, validation) #8730

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1693b38
feat(nodepool): integrate osImageStream into NodePool controller
sdminonne Jul 1, 2026
c97d648
fix(nodepool): hardcode boot image stream to rhel-9 until MCO support…
sdminonne Jul 3, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions api/hypershift/v1beta1/nodepool_types.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -264,8 +264,10 @@ type NodePoolSpec struct {

const (
// OSImageStreamRHEL9 is the OS image stream name for RHEL 9.
// This value must match support/releaseinfo.StreamRHEL9.
OSImageStreamRHEL9 = "rhel-9"
// OSImageStreamRHEL10 is the OS image stream name for RHEL 10.
// This value must match support/releaseinfo.StreamRHEL10.
OSImageStreamRHEL10 = "rhel-10"
)

Expand Down
18 changes: 9 additions & 9 deletions hypershift-operator/controllers/nodepool/aws.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,8 +60,8 @@ func isSpotEnabled(nodePool *hyperv1.NodePool) bool {
return false
}

func awsMachineTemplateSpec(infraName string, hostedCluster *hyperv1.HostedCluster, nodePool *hyperv1.NodePool, defaultSG bool, releaseImage *releaseinfo.ReleaseImage) (*capiaws.AWSMachineTemplateSpec, error) {
ami, err := resolveAWSAMI(hostedCluster, nodePool, releaseImage)
func awsMachineTemplateSpec(infraName string, hostedCluster *hyperv1.HostedCluster, nodePool *hyperv1.NodePool, defaultSG bool, releaseImage *releaseinfo.ReleaseImage, rhelStream string) (*capiaws.AWSMachineTemplateSpec, error) {
ami, err := resolveAWSAMI(hostedCluster, nodePool, releaseImage, rhelStream)
if err != nil {
return nil, err
}
Expand Down Expand Up @@ -118,7 +118,7 @@ func awsMachineTemplateSpec(infraName string, hostedCluster *hyperv1.HostedClust
return awsMachineTemplateSpec, nil
}

func resolveAWSAMI(hostedCluster *hyperv1.HostedCluster, nodePool *hyperv1.NodePool, releaseImage *releaseinfo.ReleaseImage) (string, error) {
func resolveAWSAMI(hostedCluster *hyperv1.HostedCluster, nodePool *hyperv1.NodePool, releaseImage *releaseinfo.ReleaseImage, rhelStream string) (string, error) {
// TODO: Should the region be included in the NodePool platform information?
region := hostedCluster.Spec.Platform.AWS.Region
arch := nodePool.Spec.Arch
Expand All @@ -134,8 +134,7 @@ func resolveAWSAMI(hostedCluster *hyperv1.HostedCluster, nodePool *hyperv1.NodeP
return ami, nil
}
// Default behavior for Linux/RHCOS AMIs
// TODO(CNTRLPLANE-3553): resolve streamName via GetRHELStream once osImageStream API field is available
ami, err := defaultNodePoolAMI(region, arch, "", releaseImage)
ami, err := defaultNodePoolAMI(region, arch, rhelStream, releaseImage)
if err != nil {
return "", fmt.Errorf("couldn't discover an AMI for release image: %w", err)
}
Expand Down Expand Up @@ -271,7 +270,7 @@ func awsAdditionalTags(nodePool *hyperv1.NodePool, hostedCluster *hyperv1.Hosted
}

func (c *CAPI) awsMachineTemplate(ctx context.Context, templateNameGenerator func(spec any) (string, error)) (*capiaws.AWSMachineTemplate, error) {
desiredSpec, err := awsMachineTemplateSpec(c.capiClusterName, c.hostedCluster, c.nodePool, c.cpoCapabilities.CreateDefaultAWSSecurityGroup, c.releaseImage)
desiredSpec, err := awsMachineTemplateSpec(c.capiClusterName, c.hostedCluster, c.nodePool, c.cpoCapabilities.CreateDefaultAWSSecurityGroup, c.releaseImage, c.resolvedRHELStreamForBootImage)
if err != nil {
return nil, fmt.Errorf("failed to generate AWSMachineTemplateSpec: %w", err)
}
Expand Down Expand Up @@ -361,9 +360,10 @@ func (r *NodePoolReconciler) setAWSConditions(_ context.Context, nodePool *hyper
ObservedGeneration: nodePool.Generation,
})
} else {
// Default behavior for Linux/RHCOS AMIs
// TODO(CNTRLPLANE-3553): resolve streamName via GetRHELStream once osImageStream API field is available
ami, err := defaultNodePoolAMI(hcluster.Spec.Platform.AWS.Region, nodePool.Spec.Arch, "", releaseImage)
// Default behavior for Linux/RHCOS AMIs.
// TODO(CNTRLPLANE-3553): hardcode to rhel-9 until the MCO can install
// rhel-10 OS images. Use getRHELStreamForBootImage once MCO support lands.
ami, err := defaultNodePoolAMI(hcluster.Spec.Platform.AWS.Region, nodePool.Spec.Arch, StreamRHEL9, releaseImage)
if err != nil {
SetStatusCondition(&nodePool.Status.Conditions, hyperv1.NodePoolCondition{
Type: hyperv1.NodePoolValidPlatformImageType,
Expand Down
182 changes: 168 additions & 14 deletions hypershift-operator/controllers/nodepool/aws_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -298,6 +298,7 @@ func TestAWSMachineTemplateSpec(t *testing.T) {
},
true,
releaseImage,
"",
)
if tc.checkError != nil {
tc.checkError(t, err)
Expand Down Expand Up @@ -1090,19 +1091,7 @@ func TestSetAWSConditions(t *testing.T) {
ImageStream: &v1.ImageStream{
ObjectMeta: metav1.ObjectMeta{Name: "4.17.0"},
},
StreamMetadata: &stream.Stream{
Architectures: map[string]stream.Arch{
"x86_64": {
Images: stream.Images{
Aws: &stream.AwsImage{
Regions: map[string]stream.SingleImage{
"us-east-1": {Release: "4.17.0", Image: "ami-linux-us-east-1"},
},
},
},
},
},
},
StreamMetadata: testAWSStreamWithRelease("x86_64", "us-east-1", "ami-linux-us-east-1", "4.17.0"),
}

testCases := []struct {
Expand Down Expand Up @@ -1175,6 +1164,14 @@ func TestSetAWSConditions(t *testing.T) {
expectedCondType: string(hyperv1.NodePoolValidPlatformImageType),
expectedCondValue: corev1.ConditionFalse,
},
// TODO(CNTRLPLANE-3553): re-enable once getRHELStreamForBootImage is
// wired back into setAWSConditions after MCO rhel-10 support lands.
// Currently the stream is hardcoded to rhel-9 so this validation
// path is not exercised.
// {
// name: "When osImageStream is invalid for the release version it should set ValidPlatformImage to false",
// ...
// },
{
name: "When HostedCluster has no AWS platform it should return error",
nodePool: &hyperv1.NodePool{
Expand Down Expand Up @@ -1250,6 +1247,7 @@ func TestResolveAWSAMI(t *testing.T) {
hostedCluster *hyperv1.HostedCluster
nodePool *hyperv1.NodePool
releaseImage *releaseinfo.ReleaseImage
rhelStream string
expectedAMI string
expectError bool
}{
Expand Down Expand Up @@ -1335,12 +1333,100 @@ func TestResolveAWSAMI(t *testing.T) {
},
expectError: true,
},
{
name: "When rhelStream is rhel-9 with single-stream payload (OSStreams nil), it should fall back to StreamMetadata",

@enxebre enxebre Jul 1, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nice test cases!

hostedCluster: &hyperv1.HostedCluster{
Spec: hyperv1.HostedClusterSpec{
Platform: hyperv1.PlatformSpec{AWS: &hyperv1.AWSPlatformSpec{Region: "us-east-1"}},
},
},
nodePool: &hyperv1.NodePool{
Spec: hyperv1.NodePoolSpec{
Arch: hyperv1.ArchitectureAMD64,
Platform: hyperv1.NodePoolPlatform{AWS: &hyperv1.AWSNodePoolPlatform{}},
},
},
rhelStream: "rhel-9",
releaseImage: &releaseinfo.ReleaseImage{
ImageStream: &v1.ImageStream{ObjectMeta: metav1.ObjectMeta{Name: "4.17.0"}},
StreamMetadata: testAWSStreamWithRelease("x86_64", "us-east-1", "ami-fallback-stream-metadata", "4.17.0"),
OSStreams: nil,
},
expectedAMI: "ami-fallback-stream-metadata",
},
{
name: "When rhelStream is rhel-9 with multi-stream payload, it should use OSStreams rhel-9",
hostedCluster: &hyperv1.HostedCluster{
Spec: hyperv1.HostedClusterSpec{
Platform: hyperv1.PlatformSpec{AWS: &hyperv1.AWSPlatformSpec{Region: "us-east-1"}},
},
},
nodePool: &hyperv1.NodePool{
Spec: hyperv1.NodePoolSpec{
Arch: hyperv1.ArchitectureAMD64,
Platform: hyperv1.NodePoolPlatform{AWS: &hyperv1.AWSNodePoolPlatform{}},
},
},
rhelStream: "rhel-9",
releaseImage: &releaseinfo.ReleaseImage{
ImageStream: &v1.ImageStream{ObjectMeta: metav1.ObjectMeta{Name: "5.0.0"}},
StreamMetadata: testAWSStreamWithRelease("x86_64", "us-east-1", "ami-default-stream", "5.0.0"),
OSStreams: map[string]*stream.Stream{
"rhel-9": testAWSStreamWithRelease("x86_64", "us-east-1", "ami-rhel9-osstreams", "5.0.0"),
},
},
expectedAMI: "ami-rhel9-osstreams",
},
{
name: "When rhelStream is rhel-10 with multi-stream payload, it should use OSStreams rhel-10",
hostedCluster: &hyperv1.HostedCluster{
Spec: hyperv1.HostedClusterSpec{
Platform: hyperv1.PlatformSpec{AWS: &hyperv1.AWSPlatformSpec{Region: "us-east-1"}},
},
},
nodePool: &hyperv1.NodePool{
Spec: hyperv1.NodePoolSpec{
Arch: hyperv1.ArchitectureAMD64,
Platform: hyperv1.NodePoolPlatform{AWS: &hyperv1.AWSNodePoolPlatform{}},
},
},
rhelStream: "rhel-10",
releaseImage: &releaseinfo.ReleaseImage{
ImageStream: &v1.ImageStream{ObjectMeta: metav1.ObjectMeta{Name: "5.0.0"}},
StreamMetadata: testAWSStreamWithRelease("x86_64", "us-east-1", "ami-default-stream", "5.0.0"),
OSStreams: map[string]*stream.Stream{
"rhel-10": testAWSStreamWithRelease("x86_64", "us-east-1", "ami-rhel10-osstreams", "5.0.0"),
},
},
expectedAMI: "ami-rhel10-osstreams",
},
{
name: "When rhelStream is rhel-10 with single-stream payload (OSStreams nil), it should fall back to StreamMetadata",
hostedCluster: &hyperv1.HostedCluster{
Spec: hyperv1.HostedClusterSpec{
Platform: hyperv1.PlatformSpec{AWS: &hyperv1.AWSPlatformSpec{Region: "us-east-1"}},
},
},
nodePool: &hyperv1.NodePool{
Spec: hyperv1.NodePoolSpec{
Arch: hyperv1.ArchitectureAMD64,
Platform: hyperv1.NodePoolPlatform{AWS: &hyperv1.AWSNodePoolPlatform{}},
},
},
rhelStream: "rhel-10",
releaseImage: &releaseinfo.ReleaseImage{
ImageStream: &v1.ImageStream{ObjectMeta: metav1.ObjectMeta{Name: "4.18.0"}},
StreamMetadata: testAWSStreamWithRelease("x86_64", "us-east-1", "ami-legacy-fallback", "4.18.0"),
OSStreams: nil,
},
expectedAMI: "ami-legacy-fallback",
},
}

for _, tc := range testCases {
t.Run(tc.name, func(t *testing.T) {
g := NewWithT(t)
ami, err := resolveAWSAMI(tc.hostedCluster, tc.nodePool, tc.releaseImage)
ami, err := resolveAWSAMI(tc.hostedCluster, tc.nodePool, tc.releaseImage, tc.rhelStream)
if tc.expectError {
g.Expect(err).To(HaveOccurred())
} else {
Expand Down Expand Up @@ -1864,3 +1950,71 @@ func TestApplyAWSPlacementOptions(t *testing.T) {
})
}
}

// TestAWSMachineTemplateSpec_StreamSelection verifies that on a multi-stream
// OCP 5.0+ payload, passing different rhelStream values to awsMachineTemplateSpec
// selects different AMIs and produces different machine template name hashes,
// confirming that a stream switch triggers a CAPI node rollout.
func TestAWSMachineTemplateSpec_StreamSelection(t *testing.T) {
g := NewWithT(t)

const (
legacyAMI = "ami-legacy-rhel9"
rhel10AMI = "ami-rhel10-new"
region = "us-east-1"
)

releaseImage := &releaseinfo.ReleaseImage{
ImageStream: &v1.ImageStream{ObjectMeta: metav1.ObjectMeta{Name: "5.0.0"}},
StreamMetadata: testAWSStreamWithRelease("x86_64", region, legacyAMI, "5.0.0"),
OSStreams: map[string]*stream.Stream{
"rhel-9": testAWSStreamWithRelease("x86_64", region, legacyAMI, "5.0.0"),
"rhel-10": testAWSStreamWithRelease("x86_64", region, rhel10AMI, "5.0.0"),
},
}

hostedCluster := &hyperv1.HostedCluster{
Spec: hyperv1.HostedClusterSpec{
Platform: hyperv1.PlatformSpec{
AWS: &hyperv1.AWSPlatformSpec{Region: region},
},
},
Status: hyperv1.HostedClusterStatus{
Platform: &hyperv1.PlatformStatus{
AWS: &hyperv1.AWSPlatformStatus{DefaultWorkerSecurityGroupID: "sg-default"},
},
},
}

nodePool := &hyperv1.NodePool{
ObjectMeta: metav1.ObjectMeta{Name: "test-nodepool"},
Spec: hyperv1.NodePoolSpec{
Arch: hyperv1.ArchitectureAMD64,
Platform: hyperv1.NodePoolPlatform{
Type: hyperv1.AWSPlatform,
AWS: &hyperv1.AWSNodePoolPlatform{},
},
},
}

// Legacy path (empty stream) should select the RHEL-9 AMI from StreamMetadata.
legacySpec, err := awsMachineTemplateSpec(infraName, hostedCluster, nodePool, true, releaseImage, "")
g.Expect(err).ToNot(HaveOccurred())
g.Expect(*legacySpec.Template.Spec.AMI.ID).To(Equal(legacyAMI))

// Concrete "rhel-10" stream should select the RHEL-10 AMI from OSStreams.
rhel10Spec, err := awsMachineTemplateSpec(infraName, hostedCluster, nodePool, true, releaseImage, "rhel-10")
g.Expect(err).ToNot(HaveOccurred())
g.Expect(*rhel10Spec.Template.Spec.AMI.ID).To(Equal(rhel10AMI))

// Different AMIs should produce different machine template name hashes,
// causing CAPI to create a new infrastructure ref and trigger node replacement.
legacyJSON, err := json.Marshal(legacySpec)
g.Expect(err).ToNot(HaveOccurred())
rhel10JSON, err := json.Marshal(rhel10Spec)
g.Expect(err).ToNot(HaveOccurred())

g.Expect(generateMachineTemplateName(nodePool, legacyJSON)).
ToNot(Equal(generateMachineTemplateName(nodePool, rhel10JSON)),
"different streams should produce different machine template names")
}
2 changes: 2 additions & 0 deletions hypershift-operator/controllers/nodepool/azure.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,6 +122,8 @@ func defaultAzureNodePoolImage(nodePool *hyperv1.NodePool, releaseImage *release

// getAzureMarketplaceMetadata extracts Azure Marketplace metadata from the release payload
func getAzureMarketplaceMetadata(releaseImage *releaseinfo.ReleaseImage, arch string) (*azureMarketplaceMetadata, error) {
// TODO(CNTRLPLANE-3553): use releaseImage.StreamForName(rhelStream) instead of
// accessing StreamMetadata directly, to support dual-stream payloads.
if releaseImage.StreamMetadata == nil {
return nil, nil // No stream metadata available
}
Expand Down
15 changes: 15 additions & 0 deletions hypershift-operator/controllers/nodepool/conditions.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -368,6 +368,20 @@ func (r *NodePoolReconciler) validMachineConfigCondition(ctx context.Context, no
return &ctrl.Result{}, nil
}

// Validate osImageStream before expensive config generation to fail fast.
// TODO(CNTRLPLANE-3553): add integration test covering this condition path
// (invalid osImageStream.Name → ValidMachineConfig condition False + error return).
if err := validateOSImageStream(nodePool, releaseImage); err != nil {
Comment thread
sdminonne marked this conversation as resolved.
SetStatusCondition(&nodePool.Status.Conditions, hyperv1.NodePoolCondition{
Type: hyperv1.NodePoolValidMachineConfigConditionType,
Status: corev1.ConditionFalse,
Reason: hyperv1.NodePoolValidationFailedReason,
Message: err.Error(),
ObservedGeneration: nodePool.Generation,
})
return &ctrl.Result{}, fmt.Errorf("failed to validate osImageStream: %w", err)
}

@jparrill jparrill Jun 17, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This runs after `NewConfigGenerator` succeeds — meaning we do expensive MCO config generation (listing core configs, parsing MachineConfigs) before checking a simple string allowlist. Moving `validateOSImageStream` before the config generation would fail fast for invalid stream names.

Not blocking, but worth reordering.

@sdminonne sdminonne Jun 18, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done. validateOSImageStream is now called before NewConfigGenerator in conditions.go, so invalid stream names fail fast before expensive MCO config generation — see 39be8a3.

haproxyRawConfig, err := r.generateHAProxyRawConfig(ctx, nodePool, hcluster, releaseImage)
if err != nil {
return &ctrl.Result{}, fmt.Errorf("failed to generate HAProxy raw config: %w", err)
Expand All @@ -385,6 +399,7 @@ func (r *NodePoolReconciler) validMachineConfigCondition(ctx context.Context, no
})
return &ctrl.Result{}, fmt.Errorf("failed to generate config: %w", err)
}

SetStatusCondition(&nodePool.Status.Conditions, hyperv1.NodePoolCondition{
Type: hyperv1.NodePoolValidMachineConfigConditionType,
Status: corev1.ConditionTrue,
Expand Down
Loading