Download a model into PVC

1. First create PVC in the namespace (‘Data Science Project’) where you want to use ‘Data Connection’ type URI.

kubectl apply -f -
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: granite-8b-code-base-pvc
spec:
  accessModes:
    - ReadWriteMany
  volumeMode: Filesystem
  resources:
    requests:
      storage: <model size>
  storageClassName: <storage class>

2. Next, create the downloader pod and update the environment variables to configure the object storage connection. This pod will download the model artifacts into the attached PVC.

apiVersion: v1
kind: Pod
metadata:
  name: granite-8b-instruct
  labels:
    name: granite-8b-instruct
spec:
  volumes:
    - name: model-volume
      persistentVolumeClaim:
        claimName: granite-8b-code-claim
  restartPolicy: Never
  initContainers:
    - name: fix-volume-permissions
      image: quay.io/quay/busybox@sha256:92f3298bf80a1ba949140d77987f5de081f010337880cd771f7e7fc928f8c74d
      command: ["sh"]
      args: ["-c", "mkdir -p /mnt/models/$(MODEL_PATH) && chmod -R 777 /mnt/models"] 
      volumeMounts:
        - mountPath: "/mnt/models/"
          name: model-volume
      env:
        - name: MODEL_PATH
          value: <model path> 
  containers:
    - resources:
        requests:
          memory: 40Gi
      name: download-model
      imagePullPolicy: IfNotPresent
      image: quay.io/opendatahub/kserve-storage-initializer:v0.14 
      args:
        - 's3://$(BUCKET_NAME)/$(MODEL_PATH)/'
        - /mnt/models/$(MODEL_PATH)
      env:
        - name: AWS_ACCESS_KEY_ID
          value: <Add real value Here>
        - name: AWS_SECRET_ACCESS_KEY
          value: <Add real value Here>
        - name: BUCKET_NAME
          value: prasad 
        - name: MODEL_PATH
          value: /ibm-granite/granite-3.0-8b-instruct/ 
        - name: S3_USE_HTTPS
          value: "1"
        - name: AWS_ENDPOINT_URL
          value: https://hel1.your-objectstorage.com
        - name: awsAnonymousCredential
          value: 'false'
        - name: AWS_DEFAULT_REGION
          value: none
        - name: S3_VERIFY_SSL
          value: 'true' 
      volumeMounts:
        - mountPath: "/mnt/models/"
          name: model-volume

Now in OpenShift AI, create new storage connection with below configuration.

  storageUri: pvc://<pvc name>/<model path>

Use this storage connection to skip storage initialisation time.

Note : the storageUri should be this: pvc://<model name>. No need to put /mnt/model in the path.