databricks-unity-catalog

Query the system.access.table_lineage table filtering by target_table_full_name to find upstream sources, or by source_table_full_name to find downstream consumers. Always filter on event_date or event_time to limit the scan, since lineage data is retained for 365 days.

Use the upload_to_volume MCP tool with a local path and a /Volumes/catalog/schema/volume/ destination, or the Python SDK's w.files.upload_from method with use_parallel=True for large files. You need WRITE VOLUME privilege plus USE CATALOG and USE SCHEMA on the parents.

Query system.billing.usage grouped by usage_date, sku_name, or workspace_id to compute DBU consumption. Join with system.billing.list_prices on sku_name and cloud to estimate costs, filtering price_end_time IS NULL for current prices.

PERMISSION_DENIED means you lack READ VOLUME or WRITE VOLUME on the volume, or USE CATALOG/USE SCHEMA on its parents. Ask an admin to grant the required privilege, for example GRANT READ VOLUME ON VOLUME main.default.my_volume TO your_group.

Audit logs, billing usage, and lineage data are retained for 365 days, while query history and compute events are retained for 30 days. Plan scheduled aggregation jobs if you need longer retention for compliance or trending.

System schemas must be enabled per metastore before querying. Use the Python SDK w.system_schemas.enable with your metastore ID and schema name such as access or billing, or run databricks system-schemas enable from the CLI.