Restore Inference Servers¶

DataRobot uses inference servers as part of its serverless predictions offering. If the associated serverless resources are not restored, customers will be unable to make predictions using model deployments created prior to the backup.

To ensure inference servers are properly restored, you must include the following Kubernetes Custom Resource Definitions (CRDs) in your restore procedure:

• executionenvironments.predictions.datarobot.com
• inferenceservers.predictions.datarobot.com

Prerequisites¶

• Utility kubectl of version 1.23 is installed on the host where backup will be created

• Utility kubectl is configured to access the Kubernetes cluster where DataRobot application is running, verify this with kubectl cluster-info command.

Restore Kubernetes CRDs¶

Assuming the desired Kubernetes namespace and the backup location environment variables are set:

export DR_CORE_NAMESPACE=dr11p1
export BACKUP_LOCATION=~/datarobot-backups/

The Kubernetes CRDs can be restored using the following script:

echo "executionenvironments.predictions.datarobot.com restore started"
for ee in $(ls $BACKUP_LOCATION/predictions/executionenvironments); do
    echo $ee;
    kubectl apply -n $DR_CORE_NAMESPACE -f $BACKUP_LOCATION/predictions/executionenvironments/$ee;
done
echo "executionenvironments.predictions.datarobot.com restore finished"


echo "inferenceservers.predictions.datarobot.com restore started"
for is in $(ls $BACKUP_LOCATION/predictions/inferenceservers); do
    echo $is;
    kubectl apply -n $DR_CORE_NAMESPACE -f $BACKUP_LOCATION/predictions/inferenceservers/$is
done
echo "inferenceservers.predictions.datarobot.com restore finished"