Verification checks

Before anything is deployed, the installer runs a pre-flight against your machine and your workspace. The point is simple: an install that cannot succeed should fail here, where nothing has been created yet, rather than half-way through a deploy.

Installer, “Workspace verification” panel with every check passed and the Deploy button enabled.

Select Re-run checks after fixing anything. You do not need to restart the installer.

The checks

Local

Check	Blocking	What it verifies
Databricks CLI installed	Yes	The `databricks` command is on your PATH. The deploy shells out to it.

This is the packaged installer’s only hard local prerequisite.

Workspace

Check	Blocking	What it verifies
Workspace authentication	Yes	The cached credential works against the hub workspace.
Catalog exists and is accessible	Yes	The catalog you chose is visible to the identity you signed in as.
SQL warehouse reachable	Yes	The warehouse you chose exists and can be read.
Run-as service principal (`lakemine-scan`)	Yes	Service principals can be listed.
*`system.` schemas enabled**	No	The Databricks system schemas are available.

The four blocking checks must pass before Deploy turns on.

The service principal check passing does not mean lakemine-scan already exists. Absent is fine, and the deploy creates it. What is being verified is that service principals can be listed at all, because the create-or-reuse step would fail the same way mid-deploy if they could not.

The `system.*` check is a warning, not a blocker

This is the one check that can fail and still let you deploy, and it is deliberate.

Four schemas are required for full coverage:

Schema	What it supplies	Without it
`system.billing`	Spend and DBU data	No dollars. This is the cost floor.
`system.compute`	Cluster inventory, node types, utilization	Cluster sizing and utilization detections skip.
`system.lakeflow`	Jobs, tasks, runs, pipelines	Job and pipeline detections skip.
`system.query`	SQL statement history	Query and warehouse detections skip.

One is optional:

Schema	What it supplies	Without it
`system.access`	Workspace names	The dashboard shows workspace ids instead. Automatic fallback, nothing else affected.

Lakemine never silently half-works. If a schema is missing, the detections that need it are skipped and named, both in the installer’s log and in the scan job’s output, and everything else runs.

Installer, “Workspace verification” panel with the system.* check showing a warning rather than a failure, naming the schemas that are unavailable and the detections that will be skipped, with Deploy still enabled.

Clearing a failure

Databricks CLI installed. Install it and select Re-run checks. The installer shows the one-line command for your operating system next to the failed check:


# macOS
brew tap databricks/tap && brew install databricks
 
# Windows
winget install Databricks.DatabricksCLI
 
# Linux
curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sudo sh

Workspace authentication. The cached workspace sign-in has expired or was never completed. Go back to Choose the hub workspace and sign in to the workspace again.

Catalog exists and is accessible. Either the catalog name is wrong, or the identity you signed in as cannot see it. Go back to Deployment destination, select Reload catalogs, and pick one from the list rather than typing it.

SQL warehouse reachable. The warehouse was deleted or stopped being visible to you since you picked it. Reload the warehouse list and pick again, or let the installer create a dedicated lakemine-wh.

Run-as service principal. You cannot list service principals in this workspace, which usually means the identity you signed in as is not a workspace admin. Sign in as a workspace admin, or have one run the install.

system.* schemas enabled. Enable the system schemas in your Databricks account. This is an account-level action, documented by Databricks. You can deploy without it, but with no system.billing there are no dollars to report, so it is worth clearing before you deploy rather than after.

What has been created at this point

Nothing. The pre-flight is entirely read-only. If you close the installer here, your Databricks account is exactly as it was.

Deploy and first scan.