Skip to main content

WF-004: Workstation Provisioning Workflow

Overview​

This workflow automatically provisions GCP Cloud Workstations when a user completes subscription checkout. It determines the appropriate machine configuration based on subscription tier and handles the full lifecycle from creation to user notification.

Trigger: Pub/Sub message on workstation-provisioning-sub Duration: ~30-120 seconds (GCP workstation creation) Related Workflows: WF-002 (Checkout), WF-003 (Stripe Webhook)

Prerequisites​

Before starting, ensure you have:

  • Required tools installed
  • Access to necessary resources
  • Basic understanding of concepts

Verify setup:

# Verification command

Workflow Diagram​

Workstation Provisioning Flow

Step-by-Step Narrative​

Step 1: Provisioning Event Received​

  • Node: Provision Event Trigger
  • Type: Google Cloud Pub/Sub Trigger
  • Subscription: workstation-provisioning-sub
  • Actions:
    • Receives provisioning request from Pub/Sub
    • Extracts message data: user_id, org_id, tier
    • Acknowledges message to prevent redelivery

Step 2: Determine Workstation Configuration​

  • Node: Determine Workstation Config
  • Type: Code (JavaScript)
  • Actions:
    • Maps subscription tier to machine configuration:
      • Starter: e2-medium, 50GB disk
      • Professional: e2-highmem-4, 100GB disk
      • Business: n2-highmem-8, 200GB disk
      • Enterprise: n2-highmem-16, 500GB disk
    • Generates unique workstation ID: ws-{user_id_prefix}
    • Prepares configuration payload

Step 3: Create Workstation Record​

  • Node: Create Workstation Record
  • Type: PostgreSQL Insert
  • Table: public.workstations
  • Actions:
    • Inserts workstation record with status creating
    • Stores machine type and disk size
    • Associates with user and organization
    • Records creation timestamp

Step 4: Call GCP Workstations API​

  • Node: Create GCP Workstation
  • Type: HTTP Request (REST API)
  • Endpoint: workstations.googleapis.com/v1/.../workstations
  • Method: POST
  • Actions:
    • Creates workstation in GCP
    • Specifies workstation config based on tier
    • Sets environment variables for user context
    • Returns operation name for status polling

Step 5: Wait for Creation​

  • Node: Wait for Creation
  • Type: Wait Node
  • Duration: 10 seconds
  • Actions:
    • Pauses workflow to allow GCP to create workstation
    • Prevents excessive API polling
    • Allows GCP time to initialize resources

Step 6: Check Workstation Status​

  • Node: Check Workstation Status
  • Type: HTTP Request (REST API)
  • Endpoint: workstations.googleapis.com/v1/{workstation_name}
  • Method: GET
  • Actions:
    • Polls workstation status from GCP
    • Checks for STATE_RUNNING
    • Returns current state and metadata

Step 7: Status Check Loop​

  • Node: Is Running?
  • Type: If/Else Condition
  • Condition: state === 'STATE_RUNNING'
  • Actions:
    • If Running: Proceed to update and notify
    • If Not Running: Loop back to wait and check again
    • Maximum 12 iterations (2 minutes total)

Step 8: Update Status to Running​

  • Node: Update Status Running
  • Type: PostgreSQL Update
  • Table: public.workstations
  • Actions:
    • Updates status to running
    • Stores GCP workstation resource name
    • Records provisioned_at timestamp
    • Updates organization workstation_count

Step 9: Send Ready Email​

  • Node: Send Ready Email
  • Type: Email Send
  • Actions:
    • Notifies user their workstation is ready
    • Includes machine specifications
    • Provides "Launch Workstation" button/link
    • Sent from: noreply@coditect.ai

Step 10: Publish Ready Event​

  • Node: Publish Ready Event
  • Type: Google Cloud Pub/Sub
  • Topic: workstation-events
  • Actions:
    • Publishes workstation.ready event
    • Includes workstation ID and user ID
    • Enables analytics and monitoring integrations
    • Triggers any downstream automation

Data Flow​

Input (Pub/Sub message):
{
  "user_id": "user-uuid",
  "org_id": "org-uuid",
  "tier": "professional",
  "user_email": "user@example.com"
}

Database Record:
{
  "id": "ws-uuid",
  "organization_id": "org-uuid",
  "user_id": "user-uuid",
  "machine_type": "e2-highmem-4",
  "boot_disk_size_gb": 100,
  "status": "running",
  "gcp_workstation_id": "projects/.../workstations/ws-xxx"
}

Output (Pub/Sub event):
{
  "event": "workstation.ready",
  "user_id": "user-uuid",
  "workstation_id": "projects/.../workstations/ws-xxx",
  "timestamp": "2024-01-15T10:30:00Z"
}

Tier Configuration Matrix​

TierMachine TypevCPUsMemoryDiskMonthly Cost
Startere2-medium24GB50GB~$25
Professionale2-highmem-4432GB100GB~$80
Businessn2-highmem-8864GB200GB~$200
Enterprisen2-highmem-1616128GB500GB~$500

Error Handling​

ErrorCauseAction
GCP API 403Quota exceededAlert ops, queue for retry
GCP API 409Name conflictGenerate new ID, retry
Creation timeoutGCP delayExtend polling, alert if >5min
DB write failureConnection issueRetry with exponential backoff

Monitoring and Alerts​

  • Prometheus Metrics:

    • workstation_provisioning_duration_seconds
    • workstation_provisioning_success_total
    • workstation_provisioning_failure_total

  • Alerting Rules:

    • Provisioning takes >3 minutes → Warning
    • Provisioning fails → Critical, page on-call

Security Considerations​

  • GCP Workstation runs in tenant-isolated VPC
  • User environment variables encrypted in transit
  • Workstation ID not predictable (UUID-based)
  • OAuth2 authentication for GCP API calls
  • Audit log entry for compliance

Troubleshooting​

Common Issue 1​

Problem: Description of issue Solution: Steps to resolve

Common Issue 2​

Problem: Description of issue Solution: Steps to resolve

Next Steps​

After completing this guide:

  1. Explore: Additional related features
  2. Practice: Apply concepts in your project
  3. Reference: Related documentation