#VirtualMachineClusterInstancetype [virtualmachineclusterinstancetypes.instancetype.kubevirt.io/v1beta1]

APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources

Kind is a string value representing the REST resource this object represents. Servers may infer this from the endpoint the client submits requests to. Cannot be updated. In CamelCase. More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds

ObjectMeta is metadata that all persisted resources must have, which includes all objects users must create.

Required spec describing the instancetype

Optionally defines the required Annotations to be used by the instance type and applied to the VirtualMachineInstance

Required CPU related attributes of the instancetype.

Optionally defines any GPU devices associated with the instancetype.

Optionally defines any HostDevices associated with the instancetype.

Optionally specifies the IOThreads options to be used by the instancetype.

Optionally defines the IOThreadsPolicy to be used by the instancetype.

Optionally defines the LaunchSecurity to be used by the instancetype.

Required Memory related attributes of the instancetype.

NodeSelector is a selector which must be true for the vmi to fit on a node. Selector which must match a node's labels for the vmi to be scheduled on that node. More info: https://kubernetes.io/docs/concepts/configuration/assign-pod-node/

NodeSelector is the name of the custom node selector for the instancetype.

If specified, the VMI will be dispatched by specified scheduler. If not specified, the VMI will be dispatched by default scheduler.

SchedulerName is the name of the custom K8s scheduler for the instancetype.

DedicatedCPUPlacement requests the scheduler to place the VirtualMachineInstance on a node with enough dedicated pCPUs and pin the vCPUs to it.

Required number of vCPUs to expose to the guest.

The resulting CPU topology being derived from the optional PreferredCPUTopology attribute of CPUPreferences that itself defaults to PreferSockets.

IsolateEmulatorThread requests one more dedicated pCPU to be allocated for the VMI to place the emulator thread on it.

MaxSockets specifies the maximum amount of sockets that can be hotplugged

Model specifies the CPU model inside the VMI. List of available models https://github.com/libvirt/libvirt/tree/master/src/cpu_map. It is possible to specify special cases like "host-passthrough" to get the same CPU as the node and "host-model" to get CPU closest to the node one. Defaults to host-model.

NUMA allows specifying settings for the guest NUMA topology

Realtime instructs the virt-launcher to tune the VMI for lower latency, optional for real time workloads

GuestMappingPassthrough will create an efficient guest topology based on host CPUs exclusively assigned to a pod. The created topology ensures that memory and CPUs on the virtual numa nodes never cross boundaries of host numa nodes.

Mask defines the vcpu mask expression that defines which vcpus are used for realtime. Format matches libvirt's expressions. Example: "0-3,^1","0,2,3","2-3"

ClaimName needs to be provided from the list vmi.spec.resourceClaims[].name where this device is allocated

DeviceName is the name of the device provisioned by device-plugins

Name of the GPU device as exposed by a device plugin

RequestName needs to be provided from resourceClaim.spec.devices.requests[].name where this device is requested

If specified, the virtual network interface address and its tag will be provided to the guest via config drive

Enabled determines if a display addapter backed by a vGPU should be enabled or disabled on the guest. Defaults to true.

Enables a boot framebuffer, until the guest OS loads a real GPU driver Defaults to true.

Enabled determines if the feature should be enabled or disabled on the guest. Defaults to true.

ClaimName needs to be provided from the list vmi.spec.resourceClaims[].name where this device is allocated

DeviceName is the name of the device provisioned by device-plugins

RequestName needs to be provided from resourceClaim.spec.devices.requests[].name where this device is requested

If specified, the virtual network interface address and its tag will be provided to the guest via config drive

SupplementalPoolThreadCount specifies how many iothreads are allocated for the supplementalPool policy.

AMD Secure Encrypted Virtualization (SEV).

AMD SEV-SNP flags defined by the SEV-SNP specifications.

Intel Trust Domain Extensions (TDX).

If specified, run the attestation process for a vmi.

Base64 encoded guest owner's Diffie-Hellman key.

Guest policy flags as defined in AMD SEV API specification. Note: due to security reasons it is not allowed to enable guest debugging. Therefore NoDebug flag is not exposed to users and is always true.

Base64 encoded session blob.

SEV-ES is required. Defaults to false.

Required amount of memory which is visible inside the guest OS.

Optionally enables the use of hugepages for the VirtualMachineInstance instead of regular memory.

MaxGuest allows to specify the maximum amount of memory which is visible inside the Guest OS. The delta between MaxGuest and Guest is the amount of memory that can be hot(un)plugged.

OvercommitPercent is the percentage of the guest memory which will be overcommitted. This means that the VMIs parent pod (virt-launcher) will request less physical memory by a factor specified by the OvercommitPercent. Overcommits can lead to memory exhaustion, which in turn can lead to crashes. Use carefully. Defaults to 0

PageSize specifies the hugepage size, for x86_64 architecture valid values are 1Gi and 2Mi.