Creates a libvirt cloud config's machine pool
POST/v1/cloudconfigs/libvirt/:configUid/machinePools
Creates a libvirt cloud config's machine pool
Request
Path Parameters
Cluster's cloud config uid
Header Parameters
Scope the request to the specified project uid
- application/json
Body
- Array [
- ]
- Array [
- Array [
- ]
- Array [
- ]
- ]
- Array [
- ]
cloudConfig
instanceType objectrequired
LibvirtInstanceType defines the instance configuration for a virtual machine
cpuPassthroughSpec object
Defines CPU Passthrough Spec. A not null value enables CPU Passthrough for the libvirt domain. Further cache passthrough can be enabled with the CPU passthrough spec.
Enables the CPU Passthrough for the libvirt domain
CPUSet defines cpuset for an instance which allows allocation specific set of cpus E.g cpuset="1-4,^3,6" See https://libvirt.org/formatdomain.html#cpu-allocation
gpuConfig object
GPU configuration
addresses object
Addresses is a map of PCI device entry name to its addresses. Example entry would be "11:00.0 VGA compatible controller [0300]: NVIDIA Corporation Device [10de:1eb1] (rev a1)"- > 0000_11_00_0" The address is BDF (Bus Device Function) identifier format seperated by underscores. The first 4 bits are almost always 0000. In the above example 11 is Bus, 00 is Device,0 is function. The values of these addreses are expected in hexadecimal format
DeviceModel is the model of GPU, for a given vendor, for eg., TU104GL [Tesla T4]
NumGPUs is the number of GPUs
VendorName is the GPU vendor, for eg., NVIDIA or AMD
MemoryinMB is the memory in megabytes
NumCPUs is the number of CPUs
nonRootDisksInGB object[]
NonRootDisksInGB is the list of additional disks, if required, in GB
DataStoragePool is the storage pool from which additional disks are assigned
Managed indicates if the disk is a persistent or not. By default its false indicating all disks are ephemeral.
SizeInGB is the target size in GB of the disk to be added
placements object[]required
Placements configuration Placements If defined, will replace default values defined in VsphereClusterConfig Array means one MachinePool can span across multiple vsphere compute cluster
gpuDevices object[]
GPUDevices defines an array of gpu device for a specific edge host. This will be overridden by edge host GPU devices if configured during registration.
addresses object
Addresses is a map of PCI device entry name to its addresses. Example entry would be "11:00.0 VGA compatible controller [0300]: NVIDIA Corporation Device [10de:1eb1] (rev a1)"- > 0000_11_00_0" The address is BDF (Bus Device Function) identifier format seperated by underscores. The first 4 bits are almost always 0000. In the above example 11 is Bus, 00 is Device,0 is function. The values of these addreses are expected in hexadecimal format
Model is the model of GPU, for a given vendor, for eg., TU104GL [Tesla T4]
Vendor is the GPU vendor, for eg., NVIDIA or AMD
networks object[]
NetworkName of the libvirt network where this machine will be connected
Possible values: [default, bridge]
NetworkType specifies the type of network
RootDiskInGB is the size of a vm's root disk, in GiB
XSLTemplate defines a base64-encoded raw xsl template which will be included in the machine definition
poolConfig object
Machine pool configuration for the cluster
additionalLabels object
Additional labels to be part of the machine pool
additionalTags object
AdditionalTags is an optional set of tags to add to resources managed by the provider, in addition to the ones added by default. For eg., tags for EKS nodeGroup or EKS NodegroupIAMRole
Whether this pool is for control plane
Labels for this machine pool, example: master/worker, gpu, windows
machinePoolProperties object
Machine pool specific properties
Possible values: [amd64, arm64]
Default value: amd64
Architecture type of the pool. Default value is 'amd64'
Max size of the pool, for scaling
Min size of the pool, for scaling
Minimum number of seconds a node should be Ready, before the next node is selected for repave. Applicable only for workerpools in infrastructure cluster
Size of the pool, number of nodes/machines
taints object[]
Master or worker taints
Possible values: [NoSchedule, PreferNoSchedule, NoExecute]
The taint key to be applied to a node
Time is a wrapper around time.Time which supports correct marshaling to YAML and JSON. Wrappers are provided for many of the factory methods that the time package offers.
The taint value corresponding to the taint key.
updateStrategy object
Rolling update strategy for this machine pool if not specified, will use ScaleOut
Possible values: [RollingUpdateScaleOut, RollingUpdateScaleIn]
update strategy, either ScaleOut or ScaleIn if empty, will default to RollingUpdateScaleOut
If IsControlPlane==true && useControlPlaneAsWorker==true, then will remove master taint this will not be used for worker pools
Responses
- 201
Created successfully
Response Headers
AuditUid string
Audit uid for the request
- application/json
- Schema
- Example (from schema)
Schema
{
"uid": "string"
}