LoraAdapter v1alpha1 — production-stack.vllm.ai

apiVersion string

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 string

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

metadata object

spec object

LoraAdapterSpec defines the desired state of LoraAdapter.

adapterSource object required

AdapterSource defines where to get the LoRA adapter from.

adapterName string required

AdapterName is the name of the adapter to apply.

adapterPath string

AdapterPath is the path to the LoRA adapter weights. For local sources: required, specifies the path to the adapter For remote sources: optional, will be updated by the controller with the download path

credentialsSecretRef object

CredentialsSecretRef references a secret containing storage credentials.

key string required

Key in the secret containing the value

name string required

Name of the secret

maxAdapters integer

MaxAdapters is the maximum number of adapters to load.

format: int32

pattern string

Pattern is the pattern to use for the adapter name.

repository string

Repository is the repository to get the LoRA adapter from.

type string required

Type is the type of the adapter source.

enum: local, s3, http, huggingface

baseModel string required

BaseModel is the name of the base model this adapter is for.

loraAdapterDeploymentConfig object

DeploymentConfig defines how the adapter should be deployed

algorithm string required

Algorithm specifies which placement algorithm to use.

enum: default, ordered, equalized

replicas integer

Replicas is the number of replicas that should load this adapter.

format: int32

minimum: 0

vllmApiKey object

VLLMApiKey defines the configuration for vLLM API key authentication

secretKey string required

Key in the secret containing the API key

secretName string required

Name of the secret

status object

LoraAdapterStatus defines the observed state of LoraAdapter.

conditions []object

Condition contains details for one aspect of the current state of this API Resource.

lastTransitionTime string required

LastTransitionTime is the last time the condition transitioned from one status to another.

format: date-time

message string required

Message is a human-readable message indicating details about why the current state is set.

maxLength: 32768

reason string required

Reason is a brief reason for the condition's current status.

minLength: 1

maxLength: 1024

status string required

Status is the status of the condition.

enum: True, False, Unknown

type string required

type of condition in CamelCase.

pattern: ^([a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*/)?(([A-Za-z0-9][-A-Za-z0-9_.]*)?[A-Za-z0-9])$

maxLength: 316

loadedAdapters []object

LoadedAdapters tracks the loading status of adapters and their pod assignments.

loadTime string

LoadTime is when the adapter was loaded

format: date-time

name string required

Name is the name of the adapter

path string required

Path is the path where the adapter is loaded

podAssignments object required

PodAssignments represents the pods this adapter has been assigned to

namespace string required

Namespace is the namespace of the pod

podName string required

Pod represents the pod information

status string required

Status is the status of the adapter

message string

Message provides additional information about the current phase.

observedGeneration integer

ObservedGeneration represents the .metadata.generation that the condition was set based upon.

format: int64

minimum: 0

phase string

Phase represents the current phase of the adapter deployment.