Model Storage Bucket

A recommended approach in Google Cloud is to use the Cloud Storage FUSE CSI driver.

This allows you to mount Cloud Storage buckets as a file system, avoiding the use of cloud-specific APIs.

• Follow the steps of Enable the Cloud Storage FUSE CSI driver

• Follow the steps of Create a IAM service account key and ensure the service account has read/write/create access to your Cloud Storage bucket.

• Set up the IAM policy binding for the rasa-pro-sa Kubernetes service account

  Copy

  gcloud iam service-accounts add-iam-policy-binding \

  <YOUR SERVICE ACCOUNT EMAIL> \

  --role roles/iam.workloadIdentityUser \

  --member "serviceAccount:<YOUR PROJECT_ID>.svc.id.goog[<YOUR NAMESPACE>/rasa-pro-sa]"

  This binding will allow the Kubernetes service account to act as the IAM service account.

  NOTE: You do not need to the rasa-pro-sa Kubernetes service account first. It will be created when you deploy it with the values in the next step.

• Update your values.yml file to mount your Model Storage bucket to Rasa:

  Copy

  rasa:

  endpoints:

  models:

  enabled: false

  volumes:

  - csi:

  driver: gcsfuse.csi.storage.gke.io

  readOnly: true

  volumeAttributes:

  bucketName: <YOUR BUCKET NAME>

  mountOptions: implicit-dirs,only-dir=<YOUR DIR IN YOUR BUCKET>

  name: rasa-models

  volumeMounts:

  - name: rasa-models

  mountPath: /app/models

  readOnly: true

  serviceAccount:

  # serviceAccount.create specifies whether a service account should be created.

  # set to false if you have already created the kubernetes service account

  create: true

  annotations:

  iam.gke.io/gcp-service-account: <YOUR SERVICE ACCOUNT EMAIL>

  name: "rasa-pro-sa"

  podAnnotations:

  gke-gcsfuse/volumes: "true"