How to Deploy with Helm#

Use Helm to deploy Model Registry to a Kubernetes cluster. This guide will help you:

  • Add the Helm chart repository.

  • Configure the Helm chart to match your deployment needs.

  • Deploy and verify the microservice.

Helm simplifies Kubernetes deployments by streamlining configurations and enabling easy scaling and updates. For more details, see Helm Documentation.

Prerequisites#

Before You Begin, ensure the following:

  • System Requirements: Verify that your system meets the minimum requirements.

  • Tools Installed: Install the required tools:

    • Kubernetes CLI (kubectl)

    • Helm 3 or later

  • Cluster Access: Confirm that you have access to a running Kubernetes cluster with appropriate permissions.

This guide assumes basic familiarity with Kubernetes concepts, kubectl commands, and Helm charts. If you are new to these concepts, see:

Steps to Deploy#

  1. Create directories to be used for persistent storage by the Postgres and MinIO Docker containers**

    mkdir -p /opt/intel/mr/data/mr_postgres

mkdir -p /opt/intel/mr/data/mr_minio

useradd -u 2025 mruser

chown -R mruser:mruser /opt/intel/mr/data/mr_postgres /opt/intel/mr/data/mr_minio

    • Note: The data in these directories will persist after the containers are removed. If you would like to subsequently start the containers with no pre-existing data, delete the contents in the directories before starting the containers.

  2. Pull the Helm chart from Docker Hub

    helm pull oci://registry-1.docker.io/intel/model-registry --version 1.0.3-helm

  3. Unzip the file

    tar xvf model-registry-1.0.3-helm.tgz

  4. Navigate into the directory

    cd model-registry

  5. Open the values.yaml file and enter values for the following variables:

    • MINIO_SECRET_KEY

    • POSTGRES_PASSWORD

    For more information about the supported environment variables, refer to the Environment Variables documentation.

  6. Install the helm chart

    helm install modelregistry . -n apps --create-namespace

  7. Check the status of the pods and verify the microservice is running

    kubectl get pods --namespace apps

  8. Monitor the logs for the pod associated to the model-registry to identify any issues and verify that it is operating successfully

    kubectl logs -f {{pod_name}} -n apps

Troubleshooting#

  1. Helm Chart Not Found:

    • Check if the Helm repository was added:

      helm repo list

  2. Pods Not Running:

    • Review pod logs:

      kubectl logs {{pod-name}} -n {{namespace}}

  3. Service Unreachable:

    • Confirm the service configuration:

      kubectl get svc -n {{namespace}}

Supporting Resources#