Skip to content
Open console

Enter a topic to search documentation.

    Self-host Databricks

    Deployment prerequisites, verification, and recovery for Databricks.

    Prerequisites

    You need a Kubernetes runtime you administer, Helm, network access from the provider runtime to the hub and advertised virtual-workspace endpoints, and permission to onboard this provider. Use the common provider flow to obtain the workspace kubeconfig. The examples below use a Secret named faros-provider-kubeconfig with data key kubeconfig; the generated onboarding command is authoritative for your hub.

    Install

    Terminal
    kubectl create namespace faros-provider-databricks
    kubectl --namespace faros-provider-databricks create secret generic faros-provider-kubeconfig \
      --from-file=kubeconfig=./databricks.kubeconfig
    
    helm upgrade --install databricks oci://ghcr.io/faroshq/charts/faros-databricks-provider \
      --namespace faros-provider-databricks \
      --set hub.url=https://faros.example.com \
      --set providerKubeconfig.secretName=faros-provider-kubeconfig \
      --set catalogEntry.enabled=true

    The chart README contains the matching values table and image coordinates for the deployed source version: chart instructions . Replace the placeholder kubeconfig and hub URL; do not put bearer tokens or database URLs directly in a production values file.

    Provider requirements

    Use controller.mode=required for production so Connection → Warehouse → Table dependencies gate readiness. The default bootstrap.enabled=true and bootstrap.mode=init uses the provider kubeconfig Secret and applies the CatalogEntry; use bootstrap.mode=external only when a separate operator or GitOps process owns bootstrap.

    Verify

    Run kubectl -n faros-provider-databricks get pods and inspect /readyz through the provider Service. In a test workspace, confirm Connection, Warehouse, and Table conditions, run one bounded query, and follow the Databricks quickstart .

    Recover

    For a failed rollout, first check controller events and dependency conditions. Do not switch to rest-only as a recovery shortcut unless that mode is intentional; restore the bootstrap Secret or external bootstrap resource, then verify the heartbeat and a bounded query.

    Configuration reference

    Chart values are authoritative for this source baseline. Keep the deployed image and chart versions recorded together.