@daocloud-proto/hydra 0.11.0-dev-28 → 0.11.0-dev-29

package/admin-api/model/v1alpha1/model_weights_volume.pb.ts

@@ -25,6 +25,7 @@ export type ModelWeightsVolume = {
   mountPath?: string
   creationTimestamp?: GoogleProtobufTimestamp.Timestamp
   updateTimestamp?: GoogleProtobufTimestamp.Timestamp
+  referenced?: boolean
 }
 
 export type CreateModelWeightsVolumeRequest = {

package/management-api/model/v1alpha1/model_weights_volume.pb.ts

@@ -25,6 +25,7 @@ export type WSModelWeightsVolume = {
   mountPath?: string
   creationTimestamp?: GoogleProtobufTimestamp.Timestamp
   updateTimestamp?: GoogleProtobufTimestamp.Timestamp
+  referenced?: boolean
 }
 
 export type CreateWSModelWeightsVolumeRequest = {

package/management-api/model_serving/v1alpha1/model_serving.pb.ts

@@ -8,12 +8,30 @@ import * as HydraCommonCommon from "../../../common/common.pb"
 import * as fm from "../../../fetch.pb"
 import * as GoogleProtobufTimestamp from "../../../google/protobuf/timestamp.pb"
 import * as HydraManagement_apiModelV1alpha1Model from "../../model/v1alpha1/model.pb"
+import * as HydraManagement_apiModelV1alpha1Model_weights_volume from "../../model/v1alpha1/model_weights_volume.pb"
+
+type Absent<T, K extends keyof T> = { [k in Exclude<keyof T, K>]?: undefined };
+type OneOf<T> =
+  | { [k in keyof T]?: undefined }
+  | (
+    keyof T extends infer K ?
+      (K extends string & keyof T ? { [k in K]: T[K] } & Absent<T, K>
+        : never)
+    : never);
 
 export enum PaymentMethod {
   PAYMENT_METHOD_UNSPECIFIED = "PAYMENT_METHOD_UNSPECIFIED",
   PAY_AS_YOU_GO = "PAY_AS_YOU_GO",
 }
 
+export enum InferenceRuntime {
+  INFERENCE_RUNTIME_UNSPECIFIED = "INFERENCE_RUNTIME_UNSPECIFIED",
+  VLLM = "VLLM",
+  SGLANG = "SGLANG",
+  IMAGE_GEN = "IMAGE_GEN",
+  CUSTOM = "CUSTOM",
+}
+
 export enum ModelServingStatus {
   STATUS_UNSPECIFIED = "STATUS_UNSPECIFIED",
   DEPLOYING = "DEPLOYING",
@@ -41,6 +59,46 @@ export enum WSModelServingActionRequestAction {
   STOP = "STOP",
 }
 
+export type WeightVolumeConfig = {
+  modelId?: string
+  modelTag?: string
+  storageSource?: HydraManagement_apiModelV1alpha1Model_weights_volume.StorageSourceType
+  storageRef?: string
+  subPath?: string
+  mountPath?: string
+}
+
+
+/* hydra modified */ export type BaseResourceConfig = {
+  cpu?: number
+  memory?: number
+  gpuType?: HydraManagement_apiModelV1alpha1Model.DeployTemplateGPUType
+  gpuCount?: number
+  perGpuMemory?: number
+}
+
+export type ResourceConfig = BaseResourceConfig
+  & OneOf<{ gpuCores: number }>
+
+export type EnvVar = {
+  name?: string
+  value?: string
+}
+
+export type ImagePullSecret = {
+  username?: string
+  password?: string
+}
+
+export type RuntimeConfig = {
+  runtimeType?: InferenceRuntime
+  runtimeImage?: string
+  versionRequired?: string
+  imagePullSecret?: ImagePullSecret
+  runtimeCommand?: string
+  env?: EnvVar[]
+}
+
 export type CreateModelServingRequest = {
   name?: string
   modelId?: string
@@ -50,15 +108,23 @@ export type CreateModelServingRequest = {
   paymentMethods?: PaymentMethod
 }
 
-export type CreateWSModelServingRequest = {
+
+/* hydra modified */ export type BaseCreateWSModelServingRequest = {
   name?: string
   modelId?: string
   workspace?: number
   cluster?: string
   namespace?: string
   replicas?: number
+  nodeSize?: number
+  weightVolumeId?: string
+  resourceConfig?: ResourceConfig
+  runtimeConfig?: RuntimeConfig
 }
 
+export type CreateWSModelServingRequest = BaseCreateWSModelServingRequest
+  & OneOf<{ weightVolumeConfig: WeightVolumeConfig }>
+
 export type ModelServing = {
   id?: string
   name?: string
@@ -88,6 +154,7 @@ export type WSModelServing = {
   modelName?: string
   cluster?: string
   replicas?: number
+  nodeSize?: number
   status?: ModelServingStatus
   creationTimestamp?: GoogleProtobufTimestamp.Timestamp
   accessModelName?: string
@@ -95,6 +162,9 @@ export type WSModelServing = {
   address?: string
   modelAvatar?: string
   modelSupportFeature?: HydraManagement_apiModelV1alpha1Model.ModelSupportFeature[]
+  resourceConfig?: ResourceConfig
+  runtimeConfig?: RuntimeConfig
+  weightVolumeConfig?: WeightVolumeConfig
 }
 
 export type ListModelServingRequest = {
@@ -166,13 +236,21 @@ export type WSModelServingActionRequest = {
   action?: WSModelServingActionRequestAction
 }
 
-export type WSModelServingDeployDetectionRequest = {
+
+/* hydra modified */ export type BaseWSModelServingDeployDetectionRequest = {
   workspace?: number
   modelId?: string
   cluster?: string
   namespace?: string
+  nodeSize?: number
+  weightVolumeId?: string
+  resourceConfig?: ResourceConfig
+  runtimeConfig?: RuntimeConfig
 }
 
+export type WSModelServingDeployDetectionRequest = BaseWSModelServingDeployDetectionRequest
+  & OneOf<{ weightVolumeConfig: WeightVolumeConfig }>
+
 export type ModelServingDeployDetectionResponse = {
   ready?: boolean
   conditions?: DeployDetectCondition[]

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@daocloud-proto/hydra",
-  "version": "v0.11.0-dev-28",
+  "version": "v0.11.0-dev-29",
   "description": "",
   "author": "",
   "license": "ISC"