feat: support seeding model from Huggingface repo

Makefile

@@ -172,7 +172,7 @@ ENVTEST ?= $(LOCALBIN)/setup-envtest
 
 ## Tool Versions
 KUSTOMIZE_VERSION ?= v3.8.7
-CONTROLLER_TOOLS_VERSION ?= v0.9.2
+CONTROLLER_TOOLS_VERSION ?= v0.14.0
 
 KUSTOMIZE_INSTALL_SCRIPT ?= "https://raw.githubusercontent.com/kubernetes-sigs/kustomize/master/hack/install_kustomize.sh"
 .PHONY: kustomize

apis/resources/v1alpha1/bentorequest_types.go

@@ -88,6 +88,8 @@ type BentoRequestSpec struct {
 
 	// +kubebuilder:validation:Optional
 	DownloaderContainerEnvFrom []corev1.EnvFromSource `json:"downloaderContainerEnvFrom,omitempty"`
+	// +kubebuilder:validation:Optional
+	DownloaderContainerEnv []corev1.EnvVar `json:"downloaderContainerEnv,omitempty"`
 }
 
 // BentoRequestStatus defines the observed state of BentoRequest

bento-downloader/Dockerfile

@@ -2,6 +2,14 @@ FROM ubuntu:22.04
 
 RUN apt-get update
 RUN apt-get install -y curl jq unzip xz-utils file
+RUN apt-get update && apt-get install -y \
+    python3 \
+    python3-pip \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN pip3 install --upgrade pip
+RUN pip3 install -U "huggingface_hub[cli]"
 RUN curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip" \
     && unzip awscliv2.zip \
     && ./aws/install \

bento-downloader/Makefile

@@ -1,5 +1,5 @@
-IMAGE := quay.io/bentoml/bento-downloader:0.0.1
+IMAGE := quay.io/bentoml/bento-downloader:0.0.4
 
 build:
-	docker build -t ${IMAGE} .
+	docker buildx build --platform linux/amd64 -t ${IMAGE} .
 	docker push ${IMAGE}

config/crd/bases/resources.yatai.ai_bentos.yaml

@@ -0,0 +1,131 @@
+---
+apiVersion: apiextensions.k8s.io/v1
+kind: CustomResourceDefinition
+metadata:
+  annotations:
+    controller-gen.kubebuilder.io/version: v0.14.0
+  name: bentos.resources.yatai.ai
+spec:
+  group: resources.yatai.ai
+  names:
+    kind: Bento
+    listKind: BentoList
+    plural: bentos
+    singular: bento
+  scope: Namespaced
+  versions:
+  - additionalPrinterColumns:
+    - description: Tag
+      jsonPath: .spec.tag
+      name: Tag
+      type: string
+    - description: Image
+      jsonPath: .spec.image
+      name: Image
+      type: string
+    - jsonPath: .metadata.creationTimestamp
+      name: Age
+      type: date
+    name: v1alpha1
+    schema:
+      openAPIV3Schema:
+        description: Bento is the Schema for the bentoes API
+        properties:
+          apiVersion:
+            description: |-
+              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
+            type: string
+          kind:
+            description: |-
+              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
+            type: string
+          metadata:
+            type: object
+          spec:
+            description: BentoSpec defines the desired state of Bento
+            properties:
+              context:
+                properties:
+                  bentomlVersion:
+                    type: string
+                type: object
+              image:
+                type: string
+              imagePullSecrets:
+                items:
+                  description: |-
+                    LocalObjectReference contains enough information to let you locate the
+                    referenced object inside the same namespace.
+                  properties:
+                    name:
+                      description: |-
+                        Name of the referent.
+                        More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
+                        TODO: Add other useful fields. apiVersion, kind, uid?
+                      type: string
+                  type: object
+                  x-kubernetes-map-type: atomic
+                type: array
+              models:
+                items:
+                  properties:
+                    downloadUrl:
+                      type: string
+                    size:
+                      anyOf:
+                      - type: integer
+                      - type: string
+                      pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
+                      x-kubernetes-int-or-string: true
+                    tag:
+                      type: string
+                  required:
+                  - tag
+                  type: object
+                type: array
+              runners:
+                items:
+                  properties:
+                    modelTags:
+                      items:
+                        type: string
+                      type: array
+                    name:
+                      type: string
+                    runnableType:
+                      type: string
+                  required:
+                  - name
+                  type: object
+                type: array
+              serviceName:
+                type: string
+              tag:
+                type: string
+            required:
+            - image
+            - tag
+            type: object
+          status:
+            description: BentoStatus defines the observed state of Bento
+            properties:
+              ready:
+                description: |-
+                  INSERT ADDITIONAL STATUS FIELD - define observed state of cluster
+                  Important: Run "make" to regenerate code after modifying this file
+                type: boolean
+            required:
+            - ready
+            type: object
+        type: object
+    served: true
+    storage: true
+    subresources:
+      status: {}

config/rbac/role.yaml

@@ -2,7 +2,6 @@
 apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRole
 metadata:
-  creationTimestamp: null
   name: manager-role
 rules:
 - apiGroups:

controllers/resources/bentorequest_controller.go

@@ -1701,6 +1701,7 @@ func (r *BentoRequestReconciler) generateModelSeederPodTemplateSpec(ctx context.
 
 	model := opt.Model
 
+	modelTag := model.Tag
 	modelDownloadURL := model.DownloadURL
 	modelDownloadHeader := ""
 	if modelDownloadURL == "" {
@@ -1746,6 +1747,16 @@ func (r *BentoRequestReconciler) generateModelSeederPodTemplateSpec(ctx context.
 		}
 	}
 
+	hf_token := ""
+	if envs := opt.BentoRequest.Spec.DownloaderContainerEnv; envs != nil {
+		for _, env := range envs {
+			if env.Name == "HF_TOKEN" {
+				hf_token = env.Value
+				break
+			}
+		}
+	}
+
 	modelDirPath := "/juicefs-workspace"
 	var modelSeedCommandOutput bytes.Buffer
 	err = template.Must(template.New("script").Parse(`
@@ -1756,32 +1767,54 @@ if [ -f "{{.ModelDirPath}}/.exists" ]; then
 	exit 0
 fi
 
+if [ -n "{{.HfToken}}" ]; then
+	echo "HF_TOKEN is set, using HF_TOKEN to download model..."
+	export HF_TOKEN="{{.HfToken}}"
+fi
+
 mkdir -p {{.ModelDirPath}}
 url="{{.ModelDownloadURL}}"
-echo "Downloading model {{.ModelRepositoryName}}:{{.ModelVersion}} to /tmp/downloaded.tar..."
-if [[ ${url} == s3://* ]]; then
-	echo "Downloading from s3..."
-	aws s3 cp ${url} /tmp/downloaded.tar
-elif [[ ${url} == gs://* ]]; then
-    echo "Downloading from GCS..."
-    gsutil cp ${url} /tmp/downloaded.tar
+
+if [[ ${url} == hf://* ]]; then
+	mkdir -p /tmp/model
+	hf_url="${url:5}"
+	model_id=$(echo "${hf_url}" | cut -d '@' -f 1)
+	endpoint=$(echo "${hf_url}" | cut -d '@' -f 2)
+	revision=$(echo "{{.ModelTag}}" | cut -d ':' -f 2)
+	echo "Downloading model ${model_id} (endpoint=${endpoint}, revision=${revision}) from Huggingface..."
+	export HF_ENDPOINT=${endpoint}
+	huggingface-cli download ${model_id} --revision ${revision} --local-dir /tmp/model
+	echo "Moving model to {{.ModelDirPath}}..."
+	mv /tmp/model/* {{.ModelDirPath}}
+	rm -rf /tmp/model
 else
-	curl --fail -L -H "{{.ModelDownloadHeader}}" ${url} --output /tmp/downloaded.tar --progress-bar
+	echo "Downloading model {{.ModelRepositoryName}}:{{.ModelVersion}} to /tmp/downloaded.tar..."
+	if [[ ${url} == s3://* ]]; then
+		echo "Downloading from s3..."
+		aws s3 cp ${url} /tmp/downloaded.tar
+	elif [[ ${url} == gs://* ]]; then
+		echo "Downloading from GCS..."
+		gsutil cp ${url} /tmp/downloaded.tar
+	else
+		curl --fail -L -H "{{.ModelDownloadHeader}}" ${url} --output /tmp/downloaded.tar --progress-bar
+	fi
+	cd {{.ModelDirPath}}
+	echo "Extracting model tar file..."
+	tar -xvf /tmp/downloaded.tar
+	echo "Removing model tar file..."
+	rm /tmp/downloaded.tar
 fi
-cd {{.ModelDirPath}}
-echo "Extracting model tar file..."
-tar -xvf /tmp/downloaded.tar
 echo "Creating {{.ModelDirPath}}/.exists file..."
 touch {{.ModelDirPath}}/.exists
-echo "Removing model tar file..."
-rm /tmp/downloaded.tar
 echo "Done"
 `)).Execute(&modelSeedCommandOutput, map[string]interface{}{
 		"ModelDirPath":        modelDirPath,
 		"ModelDownloadURL":    modelDownloadURL,
 		"ModelDownloadHeader": modelDownloadHeader,
 		"ModelRepositoryName": modelRepositoryName,
 		"ModelVersion":        modelVersion,
+		"ModelTag":            modelTag,
+		"HfToken":			   hf_token,
 	})
 	if err != nil {
 		err = errors.Wrap(err, "failed to generate download command")