minilake

Check that minilake is up.

List the emulated services enabled in this minilake instance.

Wipe all in-memory state: catalogs, schemas, warehouses, jobs, runs, secrets.

Create a cluster. Transitions PENDING -> RUNNING after a short real delay.

List all clusters and their states.

Get one cluster's details and current state.

Start a terminated cluster (PENDING -> RUNNING).

Terminate a cluster (TERMINATING -> TERMINATED). The record is kept.

Get a catalog's whole structure — schemas, tables and column types — in one call.

Create a MANAGED table and fill it with exactly `rows`, creating catalog and schema if needed.

Prepare a clean working environment: catalog, schema and a warehouse.

Write a file to DBFS in one shot (base64 encoding handled for you).

Read a DBFS file's contents as text.

List a DBFS directory.

Delete a DBFS file or directory.

Create a DBFS directory, including parents.

Upload a text file to minilake's file storage.

Download a file's contents as text.

Delete a file.

List a directory's contents.

Create a directory.

Create a job definition.

List all job definitions.

Get one job definition by id.

Delete a job definition.

Trigger a job run and return its run_id immediately.

Get a run's current state and per-task results.

Get a run's real captured stdout, stderr and exit code.

List runs, optionally filtered to one job.

Request cancellation of a run.

Run Python (typically PySpark) for real and return its output.

Create a secret scope.

List all secret scopes.

Delete a secret scope and its secrets.

Store a secret.

List the secret keys in a scope (keys and timestamps only, never values).

Delete one secret.

Execute a SQL statement against minilake and return the rows.

Fetch a previously executed statement's status and results by id.

Cancel a statement.

Create a catalog. Each one becomes its own DuckDB database file, so `catalog.schema.table` addressing works natively in run_sql.

List all catalogs. None exist by default — minilake creates no default catalog.

Get one catalog by name.

Delete a catalog and detach its DuckDB database.

Create a schema inside a catalog. A `default` schema already exists in each.

List the schemas in a catalog.

Delete a schema. `full_name` is `catalog.schema`.

Create a table.

List the tables in a schema.

Get a table's columns and types. `full_name` is `catalog.schema.table`.

Check whether `catalog.schema.table` exists, without raising if it does not.

Delete a table. `full_name` is `catalog.schema.table`.

Create a volume — a real directory under the data volume for unstructured files.

List the volumes in a schema.

Delete a volume. `full_name` is `catalog.schema.volume`.

Create a SQL warehouse.

List all SQL warehouses.

Get one warehouse's details by id.

Mark a warehouse RUNNING.

Mark a warehouse STOPPED. State flag only; queries still run afterwards.

Delete a warehouse and close its DuckDB connection.

Write a notebook or script to the workspace as real bytes on disk.

Read a workspace file back as plain text.

List the objects in a workspace directory.

Get metadata (object type, language) for one workspace path.

Create a workspace directory, including parents.

Delete a workspace file or directory.