Deployment Inventory

This page lists everything the Lakemeter installer creates in your Databricks workspace. Use this as a reference for auditing or manual cleanup.

Resource Summary

Resource	Name	Type
Lakebase project	lakemeter-customer	Lakebase Autoscaling
Lakebase branch	production	Copy-on-write PostgreSQL branch
Lakebase endpoint	primary	Read-write autoscaling compute
Database	lakemeter_pricing	PostgreSQL database
Schema	lakemeter	PostgreSQL schema
Application tables	users, estimates, line_items, templates, sharing, conversation_messages, decision_records, ref_cloud_tiers, ref_workload_types	PostgreSQL tables
Stored functions	15 cost calculation functions	PostgreSQL functions
Pricing sync tables	10 tables with DBU rates, VM costs, model pricing	PostgreSQL tables
Derived reference tables	ref_fmapi_databricks_models, ref_fmapi_proprietary_models, ref_model_serving_gpu_types	PostgreSQL tables
SKU mapping	ref_sku_discount_mapping	PostgreSQL table
Secret scope	lakemeter-secrets	Databricks secret scope
Secrets	7 key-value pairs	Databricks secrets
Databricks App	lakemeter	Databricks App
App resources	7 environment variable bindings	App config
Lakebase role	App Service Principal	Database role
PostgreSQL role	lakemeter_sync_role	Password-auth role

---

Lakebase Autoscaling Project

Property	Value
Project ID	lakemeter-customer (configurable via --project-id)
Branch	production
Read-write endpoint	primary
Type	Lakebase Autoscaling
Autoscaling	1 CU – 16 CU
Scale-to-zero	After 5 minutes of inactivity
Native login	Enabled for password-auth fallback

---

Secret Scope: lakemeter-secrets

Secret Key	Description
lakebase-project	Full project resource name
lakebase-branch	Full production branch resource name
lakebase-endpoint	Full primary endpoint resource name
lakebase-host	Primary endpoint DNS host
lakebase-user	PostgreSQL role name (lakemeter_sync_role)
lakebase-database	Database name (lakemeter_pricing)
lakebase-password	Auto-generated password for lakemeter_sync_role

---

Databricks App: lakemeter

Property	Value
Name	lakemeter (configurable via --app-name)
Compute size	MEDIUM (2 vCPU, 6 GB RAM)
Runtime	Ubuntu 22.04, Python 3.11, Node.js 22.16
Source path	/Workspace/Users/{user}/apps/lakemeter
URL	https://lakemeter-<workspace-id>.<cloud>.databricksapps.com

App Resources

These environment variables are injected into the app container at runtime:

Resource Name	Environment Variable	Type	Source
lm-lakebase-project	LAKEBASE_PROJECT	Secret	lakemeter-secrets:lakebase-project
lm-lakebase-branch	LAKEBASE_BRANCH	Secret	lakemeter-secrets:lakebase-branch
lm-lakebase-endpoint	LAKEBASE_ENDPOINT	Secret	lakemeter-secrets:lakebase-endpoint
lm-db-host	DB_HOST	Secret	lakemeter-secrets:lakebase-host
lm-db-user	DB_USER	Secret	lakemeter-secrets:lakebase-user
lm-db-name	DB_NAME	Secret	lakemeter-secrets:lakebase-database
lm-claude-endpoint	CLAUDE_MODEL_ENDPOINT	Serving Endpoint	databricks-claude-opus-4-6

Service Principal

The app gets an auto-created Service Principal with:

Permission	Target	Purpose
Lakebase OAuth role	production branch	Authenticates the app Service Principal
SQL grants	lakemeter schema	CONNECT, USAGE, ALL PRIVILEGES on tables/sequences/functions
Secret READ	lakemeter-secrets scope	Read database credentials
CAN_QUERY	databricks-claude-opus-4-6	Query the Claude model endpoint

---

Cleanup

To completely remove Lakemeter from your workspace:

# 1. Delete the Databricks App
databricks apps delete lakemeter --profile <profile>

# 2. Delete the Lakebase project (destroys all branches and data)
databricks postgres delete-project projects/lakemeter-customer --profile <profile>

# 3. Delete the secrets scope
databricks secrets delete-scope lakemeter-secrets --profile <profile>

# 4. Remove bundle files (optional)
databricks workspace delete -r /Workspace/Users/{user}/.bundle/lakemeter-installer --profile <profile>

# 5. Remove app source (optional)
databricks workspace delete -r /Workspace/Users/{user}/apps/lakemeter --profile <profile>

Warning: Deleting the Lakebase project permanently destroys all branches, databases, tables, and data within it. This cannot be undone.