API Reference
Packages
tensorstack.dev/v1beta1
Package v1beta1 contains API Schema definitions for the v1beta1 API group
Resource Types
Address
Appears in:
| Field | Description |
|---|---|
url string | the service's in-cluster url, e.g. managed-simplemlservice-41309.demo.svc.cluster.local |
ports AddressPort array | an array of serivce port & protocol |
AddressPort
AddressPort stores ports defined in service of simplemlservice
Appears in:
| Field | Description |
|---|---|
port integer | The port that will be exposed by this service. |
nodePort integer | The port on each node on which this service is exposed when type=NodePort or LoadBalancer. Usually assigned by the system. If specified, it will be allocated to the service if unused or else creation of the service will fail. Default is to auto-allocate a port if the ServiceType of this Service requires one. More info: https://kubernetes.io/docs/concepts/services-networking/service/#type-nodeport |
protocol Protocol | The IP protocol for this port. Supports "TCP", "UDP", and "SCTP". Default is TCP. |
Condition
Condition for simpleMLService
Appears in:
| Field | Description |
|---|---|
lastTransitionTime Time | |
type SimpleMLServiceConditionType | |
status ConditionStatus | |
reason string | |
message string |
CustomSpec
CustomSpec defines a pod template to run model serving
Appears in:
| Field | Description |
|---|---|
spec PodSpec |
DeploymentSpec
DeploymentSpec defines the configuration for replicas & scheduler
Appears in:
| Field | Description |
|---|---|
scheduler SchedulePolicy | Scheduler responsible for handling resource allocation requests. default is default-scheduler |
replicas integer | Replicas of pods running model serving |
strategy DeploymentStrategy | The deployment strategy to use to replace existing pods with new ones |
PVCStorage
PVCStorage defines infos of pvc
Appears in:
| Field | Description |
|---|---|
name string | PVC name |
subPath string | Directory path where model is located in PVC. Must be a relative path. e.g. "model/mnist" |
mountPath string | Directory path where model locates in container. Must be absolute path, default is "/var/lib/t9k/model" |
PyTorchSpec
PyTorchSpec defines arguments for configuring PyTorch model serving
Appears in:
| Field | Description |
|---|---|
modelsFlag string | Value of torchserve's flag --models |
image string | Image of torchserve |
resources ResourceRequirements | Compute Resources required by a replica |
S3Storage
S3Storage defines infos of s3
Appears in:
| Field | Description |
|---|---|
secretRef LocalObjectReference | SecretRef is reference to the secret storing s3cmd config |
uri string | Directory path where model locates in s3. e.g. "s3://<bucket>/<dir>/" |
mountPath string | Directory path where model locates in container, default is "/var/lib/t9k/model" |
SchedulePolicy
SchedulePolicy defines which scheduler is responsible for handling resource allocation requests
Appears in:
| Field | Description |
|---|---|
t9kScheduler T9kScheduler | Use t9k-scheduler |
ServiceSpec
ServiceSpec defines the desired state of Service created by Controller
Appears in:
| Field | Description |
|---|---|
ports ServicePort array | The list of ports that are exposed by this service. More info: https://kubernetes.io/docs/concepts/services-networking/service/#virtual-ips-and-service-proxies |
type ServiceType | type determines how the Service is exposed. Defaults to ClusterIP. Valid options are ExternalName, ClusterIP, NodePort, and LoadBalancer. "ExternalName" maps to the specified externalName. "ClusterIP" allocates a cluster-internal IP address for load-balancing to endpoints. Endpoints are determined by the selector or if that is not specified, by manual construction of an Endpoints object. If clusterIP is "None", no virtual IP is allocated and the endpoints are published as a set of endpoints rather than a stable IP. "NodePort" builds on ClusterIP and allocates a port on every node which routes to the clusterIP. "LoadBalancer" builds on NodePort and creates an external load-balancer (if supported in the current cloud) which routes to the clusterIP. More info: https://kubernetes.io/docs/concepts/services-networking/service/#publishing-services-service-types |
SimpleMLService
SimpleMLService is the Schema for the simplemlservices API
Appears in:
| Field | Description |
|---|---|
apiVersion string | tensorstack.dev/v1beta1 |
kind string | SimpleMLService |
metadata ObjectMeta | Refer to Kubernetes API documentation for fields of metadata. |
spec SimpleMLServiceSpec | |
status SimpleMLServiceStatus |
SimpleMLServiceConditionType
Underlying type: string
SimpleMLServiceConditionType is a type
Appears in:
SimpleMLServiceList
SimpleMLServiceList contains a list of SimpleMLService
| Field | Description |
|---|---|
apiVersion string | tensorstack.dev/v1beta1 |
kind string | SimpleMLServiceList |
metadata ListMeta | Refer to Kubernetes API documentation for fields of metadata. |
items SimpleMLService array |
SimpleMLServiceSpec
SimpleMLServiceSpec defines the desired state of SimpleMLService
Appears in:
| Field | Description |
|---|---|
tensorflow TensorflowSpec | Spec for Tensorflow Serving (https://github.com/tensorflow/serving) |
pytorch PyTorchSpec | Spec for TorchServe |
custom CustomSpec | Custom Spec |
storage Storage | Storage of model |
DeploymentSpec DeploymentSpec | Configuration for replicas & scheduler |
service ServiceSpec | Configuration for service. Controller will create default service if spec.service not set |
SimpleMLServiceStatus
SimpleMLServiceStatus defines the observed state of SimpleMLService
Appears in:
Storage
Storage defines the storage where the model locates
Appears in:
| Field | Description |
|---|---|
s3 S3Storage | Model locates in S3 |
pvc PVCStorage | Model locates in pvc |
T9kScheduler
T9kScheduler defines SimpleMLService use t9k-scheduler.
Appears in:
| Field | Description |
|---|---|
queue string | Name of queue which SimpleMLService's pod belongs to |
TensorflowSpec
TensorflowSpec defines arguments for configuring Tensorflow model serving
Appears in:
| Field | Description |
|---|---|
image string | Image of Tensorflow Serving |
resources ResourceRequirements | Compute Resources required by a replica |