Databricks Fe Vm Workspace Deployment

This shows all cached workspaces with their URLs, creation dates, and expiration dates.

Step 2: Refresh Environment Cache (If Needed)

If no cached environments exist or you need fresh data:

python3 resources/fe_vm_client.py refresh-cache

This will:

1. Check for a valid session cookie in ~/.vibe/fe-vm/session.json
2. If no valid session, use Chrome DevTools MCP to authenticate and capture the cookie
3. Fetch all deployments from FEVM and update the cache

Step 3: Find or Deploy a Workspace

To find an existing suitable workspace:

python3 resources/environment_manager.py find --type serverless --min-days 7

This returns a workspace URL if one exists with at least 7 days remaining.

To deploy a new Serverless workspace:

python3 resources/fe_vm_client.py deploy-serverless --name my-demo --region us-east-1 --lifetime 30

To deploy a new Classic workspace:

python3 resources/fe_vm_client.py deploy-classic --name my-classic --region us-west-2 --lifetime 14

Step 4: Check Deployment Status

For new deployments, monitor the progress:

python3 resources/fe_vm_client.py status --run-id <run_id>

Or wait for completion:

python3 resources/fe_vm_client.py wait --run-id <run_id>

Step 5: Authenticate to the Workspace

Once deployed, use the databricks-authentication skill to set up CLI access:

databricks auth login <workspace_url> --profile=fe-vm-<workspace_name>

Decision Flow

Use this flow to decide what to do:

1. Check cached environments: python3 resources/environment_manager.py list
2. If suitable workspace exists (correct type, >3 days remaining): Use it
3. If no suitable workspace: Refresh cache and check FEVM: python3 resources/fe_vm_client.py refresh-cache
4. If still no suitable workspace: Deploy new one based on requirements:
   • Need Apps/Lakebase? → Deploy Serverless
   • Need custom cloud config? → Deploy Classic
   • Basic demo? → Deploy Serverless (faster)

Available Commands

environment_manager.py

fe_vm_client.py

Environment Storage

All environment data is stored in ~/.vibe/fe-vm/:

~/.vibe/fe-vm/
├── session.json          # Session cookie and expiry
├── environments.json     # Cached workspace information
└── deployments/          # Individual deployment configs
    └── <workspace_name>.json

Authentication Flow

When authentication is needed:

1. Check existing session: Look for valid cookie in ~/.vibe/fe-vm/session.json
2. If expired or missing:
   • Open Chrome DevTools MCP to navigate to FEVM
   • User completes SSO login
   • Automatically extract __Host-databricksapps cookie
   • Save to ~/.vibe/fe-vm/session.json with expiry timestamp
3. Cookie typically valid for 24-48 hours

Template Reference

AWS Regions

Available regions: us-east-1, us-east-2, us-west-1, us-west-2, eu-central-1, ap-northeast-1

Workspace and Catalog Naming

FE-VM workspaces follow a standard naming convention:

Important: Each FE-VM workspace comes with a pre-created Unity Catalog. The catalog name is derived from the workspace name by replacing hyphens with underscores and appending _catalog.

When creating schemas or tables, always use the standard catalog rather than creating a new one:

-- Use the existing catalog (example for workspace "deep-research-agent")
USE CATALOG deep_research_agent_catalog;

-- Create schemas within the standard catalog
CREATE SCHEMA IF NOT EXISTS my_demo_schema;

-- Create tables in the standard catalog
CREATE TABLE deep_research_agent_catalog.my_demo_schema.my_table (...);

This avoids permission errors and ensures consistency across the workspace.

Notes

• Always prefer reusing existing workspaces to conserve quota and deployment time
• Serverless workspaces deploy in ~5-10 minutes
• Classic workspaces deploy in ~15-20 minutes
• Maximum lifetime is 30 days