Deployment destination
Where the results go, how often the scan runs, and who can read it. This is the last screen where you make choices. Everything after it is verification and deployment.
Installer, “Deployment destination” panel, fully filled in. Workspace host, catalog and schema, scan window and schedule, SQL warehouse, dashboard viewers group, and the read-only install id at the bottom.
Workspace host
Pre-filled with the hub you picked. Edit it only if you need to point at a different host for the same workspace, for example a vanity domain.
Catalog and schema
Catalog is where the results schema is created. You need CREATE SCHEMA on it, which
is the one place the install needs write access in your estate. Select Reload catalogs
if the list looks stale after signing in.
Schema defaults to lakemine. The schema is created on the scan’s first run with
CREATE SCHEMA IF NOT EXISTS, and holds seven small result tables that the dashboard
reads.
These tables are read-only outputs. The scan rebuilds them on every run, and uninstalling drops the schema and its data. Do not point the schema at an existing schema that holds anything you want to keep.
Scan window
How far back each scan looks. The default is the last 30 days, which is the recommended setting.
| Option | When to use it |
|---|---|
| Last 7 days | A fast first look, or a workspace with very short retention. |
| Last 14 days | Shorter feedback loop while you are actively remediating. |
| Last 30 days (recommended) | The default. Long enough for weekly patterns to show up, short enough to stay cheap. |
| Last 60 days | Seasonal or monthly workloads. |
| Last 90 days | The widest view. Costs the most compute per scan. |
A longer window means a bigger query against system.* on each run, so it costs more of
your own compute. Thirty days is the balance point for most estates.
Scan schedule
How often the scan runs. The default is daily at 06:00 UTC.
| Option | Notes |
|---|---|
| Daily at 06:00 UTC (recommended) | The default. Daily rather than hourly keeps the compute you pay for modest. |
| Every 12 hours | Twice the freshness, twice the scan compute. |
| Weekly, Mondays 06:00 UTC | For estates that change slowly. |
| Paused, run manually | Nothing runs on a schedule. You trigger the lakemine-scan job yourself. |
Opening the dashboard is cheap regardless. It reads the small results tables, not
system.*.
SQL warehouse
The warehouse the scan runs its queries through. Pick an existing one, or select Create
a dedicated one and the installer provisions a small serverless warehouse named
lakemine-wh.
A dedicated warehouse makes the scan’s own cost easy to see and easy to attribute. If the installer created it, uninstalling removes it. If you picked one of your own, uninstalling leaves it alone.
Dashboard viewers group
The group whose members can open the cost dashboard. This one setting is the entire access control for your cost data.
Pick an existing group, such as your FinOps team or your workspace admins, or select Create “lakemine-viewers” for a dedicated one.
Manage access afterwards by adding and removing group members. No SQL, no grants, no
ticket to a data platform team. Viewers need no permission on the result tables, the
catalog, or system.*, because the dashboard runs its queries as the appliance’s
service principal rather than as the viewer. See
How it works.
Installer, “Deployment destination” panel with the dashboard viewers group dropdown open, showing existing workspace groups and the “Create lakemine-viewers” button beside it.
The run identity
There is nothing to configure. The scheduled scan runs as a dedicated service principal
named lakemine-scan, created automatically if it does not exist and reused on every
re-install. That is what lets an upgrade run by a different admin keep working without
re-granting anything.
Install id
Generated for you and shown read-only. It identifies this deployment, tags the rows in your results tables, and appears in the portal’s install list. Note it down if you expect to raise a support request.
What you will not find here
There is no field for a Lakemine account id or a license key. Your billing identity comes from the session you signed in with on the first screen, which is also what the subscription check and the metering credential are bound to. A field for it would only let them drift apart.