DataRobot container images¶

The DataRobot application uses Docker Hub as its primary container registry. While the DataRobot Helm chart is configured to use Docker Hub by default, it also supports the option to use a dedicated private container registry if required.

Private registry overview¶

To begin, download the main DataRobot chart.

export DOCKERHUB_USERNAME="DATAROBOT_CUSTOMER_USERNAME"
export DOCKERHUB_PASSWORD="DATAROBOT_CUSTOMER_PASSWORD"
echo $DOCKERHUB_PASSWORD | helm registry login registry-1.docker.io -u $DOCKERHUB_USERNAME --password-stdin
helm pull oci://registry-1.docker.io/datarobot/datarobot-prime --version X.X.X 

The process of loading images into your private registry is handled by a dedicated Helm plugin. To install the plugin, use the following command:

helm plugin install https://github.com/datarobot-oss/helm-datarobot-plugin.git 

To authenticate your Docker Hub session, run the following command:

echo $DOCKERHUB_PASSWORD | docker login -u $DOCKERHUB_USERNAME --password-stdin 

Container images for upgrades¶

DataRobot supports automated in-place upgrades from older versions (including 9.x and 10.x) using Kubernetes. These upgrades require transitional versions of persistent services such as MongoDB, RabbitMQ, and PostgreSQL to migrate data in persistent volumes. The corresponding images are included in the default image manifest to streamline the process.

If you are not performing a major or minor version upgrade, or if you know the exact version you are upgrading from, you can omit upgrade-only images from your download to reduce bandwidth usage and conserve space in your container registry by using the following flags with the sync, save, and images commands:

• --upgrade-from: Excludes images only needed for upgrades from versions prior to the one specified. For example, --upgrade-from=11.3 (if upgrading from DataRobot 11.3.x).
• --no-upgrade: Excludes all images needed only for upgrades (e.g., for a new installation).

Container image synchronization options¶

Option 1: Load all images using the datarobot-prime Helm chart. Replace the environment variables with your Google Cloud project values.

export PREFIX="${GCP_PROJECT_NAME}/${GAR_REPO}"
helm datarobot sync datarobot-prime-x.x.x.tgz -r "${GCP_REGION}-docker.pkg.dev" --prefix $PREFIX -u _json_key_base64 -p "${GCP_BASE64_SERVICE_ACCOUNT_KEY}" 

Option 2: Load all images using the compressed tarball artifact. Replace the environment variables with your Google Cloud project values.

export PREFIX="${GCP_PROJECT_NAME}/${GAR_REPO}"
helm datarobot load tarball-datarobot-x.x.x.tar.zst -r "${GCP_REGION}-docker.pkg.dev" --prefix $PREFIX -u _json_key_base64 -p "${GCP_BASE64_SERVICE_ACCOUNT_KEY}" 

Verify images are in a Docker registry¶

To verify that the images have loaded into your registry, you can run a curl command to pull the registry catalog. The URL format is http{s}://DOCKER_REGISTRY_URL/v2/_catalog.

The example below shows a curl command to a registry with basic authentication. Replace the placeholders with your registry's details.

curl -k -u DOCKER_REGISTRY_USERNAME:DOCKER_REGISTRY_PASSWORD https://DOCKER_REGISTRY_URL/v2/_catalog 

The response is a JSON formatted list of the images in the registry, similar to the following:

{
  "repositories": [
    "build-service",
    "datarobot/ai-application-engine",
    "datarobot/datarobot-runtime",
    "datarobot/operator-lrs",
    "image-builder",
    ...
  ]
}