Skip to content

OSAC-3182: stamp GPU fields from InstanceType onto ComputeInstance CR #281

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

Merged
Tzif-Morgen merged 1 commit into from
Aug 13, 2026
Merged

OSAC-3182: stamp GPU fields from InstanceType onto ComputeInstance CR #281

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
7 changes: 7 additions & 0 deletions ...lment-service/internal/controllers/computeinstance/computeinstance_reconciler_function.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -705,6 +705,13 @@ func (t *task) addExplicitFields(ctx context.Context, spec *osacv1alpha1.Compute
itSpec := response.GetObject().GetSpec()
spec.Cores = itSpec.GetCores()
spec.MemoryGiB = itSpec.GetMemoryGib()
if gpu := itSpec.GetGpu(); gpu != nil {

@fullsend-ai-review fullsend-ai-review Bot Aug 12, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

[low] edge-case

The nil guard passes for a proto GpuSpec present with all zero-value fields (Count=0, empty strings). CRD admission validation catches this but with a generic error. Upstream InstanceType API validation prevents this state in practice.

Suggested fix: Consider adding a validation check after the nil guard, or document that InstanceType-level validation prevents this state from reaching the reconciler.

@Tzif-Morgen Tzif-Morgen Aug 13, 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.

The reconciler's job is to stamp what the InstanceType says — it's a data copy, not a validation layer. If someone creates an InstanceType with Count=0, that's an InstanceType validation problem, not a reconciler problem.
Adding validation here would duplicate upstream validation and violate the single-responsibility pattern - the existing cores/memory stamping doesn't validate either.

spec.Gpu = &osacv1alpha1.GpuSpec{
PciDeviceSelector: gpu.GetPciDeviceSelector(),
ResourceName: gpu.GetResourceName(),
Count: gpu.GetCount(),
}
}
if ciSpec.HasRunStrategy() {
spec.RunStrategy = osacv1alpha1.RunStrategyType(ciSpec.GetRunStrategy())
}
Expand Down
138 changes: 138 additions & 0 deletions ...-service/internal/controllers/computeinstance/computeinstance_reconciler_function_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1909,6 +1909,144 @@ var _ = Describe("instance_type resolution in reconciler", func() {
Expect(spec.MemoryGiB).To(Equal(int32(8)))
})

It("resolves instance_type gpu fields onto CR spec", func() {
mockInstanceTypesClient := NewMockInstanceTypesClient(ctrl)
mockInstanceTypesClient.EXPECT().
Get(gomock.Any(), gomock.Any()).
Return(privatev1.InstanceTypesGetResponse_builder{
Object: privatev1.InstanceType_builder{
Id: "gpu-a100-8core",
Spec: privatev1.InstanceTypeSpec_builder{
Cores: 8,
MemoryGib: 64,
Gpu: privatev1.GpuSpec_builder{
PciDeviceSelector: "10DE:20B0",
ResourceName: "nvidia.com/A100",
Count: 1,
}.Build(),
}.Build(),
}.Build(),
}.Build(), nil)

t := &task{
r: &function{
logger: logger,
instanceTypesClient: mockInstanceTypesClient,
},
computeInstance: privatev1.ComputeInstance_builder{
Id: "test-gpu-instance",
Spec: privatev1.ComputeInstanceSpec_builder{
Template: &privatev1.ComputeInstanceTemplateReference{Name: "osac.templates.ocp_virt_vm"},
InstanceType: &privatev1.InstanceTypeReference{Name: "gpu-a100-8core"},
NetworkAttachments: []*privatev1.NetworkAttachment{
privatev1.NetworkAttachment_builder{Subnet: &privatev1.SubnetLocalReference{Id: subnetID}}.Build(),
},
}.Build(),
}.Build(),
hubNamespace: hubNamespace,
hubClient: fakeClient,
}

spec, err := t.buildSpec(ctx)
Expect(err).ToNot(HaveOccurred())
Expect(spec.Cores).To(Equal(int32(8)))
Expect(spec.MemoryGiB).To(Equal(int32(64)))
Expect(spec.Gpu).ToNot(BeNil())
Expect(spec.Gpu.PciDeviceSelector).To(Equal("10DE:20B0"))
Expect(spec.Gpu.ResourceName).To(Equal("nvidia.com/A100"))
Expect(spec.Gpu.Count).To(Equal(int32(1)))
})

It("leaves gpu nil when InstanceType has no gpu", func() {
mockInstanceTypesClient := NewMockInstanceTypesClient(ctrl)
mockInstanceTypesClient.EXPECT().
Get(gomock.Any(), gomock.Any()).
Return(privatev1.InstanceTypesGetResponse_builder{
Object: privatev1.InstanceType_builder{
Id: "standard-4-8",
Spec: privatev1.InstanceTypeSpec_builder{
Cores: 4,
MemoryGib: 8,
}.Build(),
}.Build(),
}.Build(), nil)

t := &task{
r: &function{
logger: logger,
instanceTypesClient: mockInstanceTypesClient,
},
computeInstance: privatev1.ComputeInstance_builder{
Id: "test-no-gpu-instance",
Spec: privatev1.ComputeInstanceSpec_builder{
Template: &privatev1.ComputeInstanceTemplateReference{Name: "osac.templates.ocp_virt_vm"},
InstanceType: &privatev1.InstanceTypeReference{Name: "standard-4-8"},
NetworkAttachments: []*privatev1.NetworkAttachment{
privatev1.NetworkAttachment_builder{Subnet: &privatev1.SubnetLocalReference{Id: subnetID}}.Build(),
},
}.Build(),
}.Build(),
hubNamespace: hubNamespace,
hubClient: fakeClient,
}

spec, err := t.buildSpec(ctx)
Expect(err).ToNot(HaveOccurred())
Expect(spec.Cores).To(Equal(int32(4)))
Expect(spec.MemoryGiB).To(Equal(int32(8)))
Expect(spec.Gpu).To(BeNil())
})

It("gpu stamping is idempotent", func() {
mockInstanceTypesClient := NewMockInstanceTypesClient(ctrl)
mockInstanceTypesClient.EXPECT().
Get(gomock.Any(), gomock.Any()).
Return(privatev1.InstanceTypesGetResponse_builder{
Object: privatev1.InstanceType_builder{
Id: "gpu-a100-8core",
Spec: privatev1.InstanceTypeSpec_builder{
Cores: 8,
MemoryGib: 64,
Gpu: privatev1.GpuSpec_builder{
PciDeviceSelector: "10DE:20B0",
ResourceName: "nvidia.com/A100",
Count: 2,
}.Build(),
}.Build(),
}.Build(),
}.Build(), nil).
Times(2)

t := &task{
r: &function{
logger: logger,
instanceTypesClient: mockInstanceTypesClient,
},
computeInstance: privatev1.ComputeInstance_builder{
Id: "test-idempotent-gpu",
Spec: privatev1.ComputeInstanceSpec_builder{
Template: &privatev1.ComputeInstanceTemplateReference{Name: "osac.templates.ocp_virt_vm"},
InstanceType: &privatev1.InstanceTypeReference{Name: "gpu-a100-8core"},
NetworkAttachments: []*privatev1.NetworkAttachment{
privatev1.NetworkAttachment_builder{Subnet: &privatev1.SubnetLocalReference{Id: subnetID}}.Build(),
},
}.Build(),
}.Build(),
hubNamespace: hubNamespace,
hubClient: fakeClient,
}

spec1, err := t.buildSpec(ctx)
Expect(err).ToNot(HaveOccurred())

spec2, err := t.buildSpec(ctx)
Expect(err).ToNot(HaveOccurred())

Expect(spec1.Gpu).ToNot(BeNil())
Expect(spec2.Gpu).ToNot(BeNil())
Expect(*spec1.Gpu).To(Equal(*spec2.Gpu))
})

It("returns error when instance_type is empty", func() {
t := &task{
r: &function{
Expand Down
Loading