Skip to content

RENDER_DEPLOYMENT.md

Latest commit

 

History

History
250 lines (184 loc) · 7.15 KB

RENDER_DEPLOYMENT.md

File metadata and controls

250 lines (184 loc) · 7.15 KB
Raw

Deploying Smart Sign Interpreter on Render

This guide walks you through deploying the Smart Sign Interpreter on Render with persistent Google Sheets storage.

Quick Start

Prerequisites

  • GitHub account with access to the repository
  • Render account (free or paid)
  • Google account
  • This completed: GOOGLE_SHEETS_SETUP.md

Step 1: Prepare Google Credentials

Before deployment, ensure you have:

  1. Google Sheets API and Drive API enabled in Google Cloud Console
  2. Service account JSON credentials file downloaded
  3. Google Sheet created and shared with the service account
  4. Spreadsheet ID copied from the URL

See GOOGLE_SHEETS_SETUP.md for detailed instructions.

Step 2: Push Code to GitHub

  1. Commit all changes:

    git add .
git commit -m "Add Google Sheets integration for persistent data storage"

  2. Push to your repository:

    git push origin main

Step 3: Create Render Service

  1. Go to render.com

  2. Sign up or log in

  3. Click "New +" → "Web Service"

  4. Connect your GitHub account (first time only)

  5. Select this repository

  6. Fill in the settings:

    • Name: smart-sign-interpreter (or your preferred name)
    • Environment: Docker
    • Region: Choose closest to your location
    • Branch: main
    • Plan: Free (or upgrade if needed)

  7. Click "Create Web Service"

Wait for Render to build and deploy (2-5 minutes).

Step 4: Configure Environment Variables

  1. In your Render service dashboard, click "Environment"

  2. Click "Add Environment Variable"

  3. Add the following variables:

    APP_HOST: 0.0.0.0
APP_PORT: 8000
MODEL_PATH: models/gesture_model.pkl
DATASET_PATH: data/datasets/gesture_dataset.csv
LOG_LEVEL: INFO
ALLOW_MISSING_MODEL: true
ENABLE_DEMO: true
GOOGLE_CREDENTIALS_PATH: [PASTE_COMPLETE_SERVICE_ACCOUNT_JSON_AS_A_SECRET]
GOOGLE_SPREADSHEET_ID: [YOUR_SPREADSHEET_ID]

    Replace [YOUR_SPREADSHEET_ID] with your actual Google Sheet ID.

Step 5: Store Google Credentials Safely

For Render free tier, create GOOGLE_CREDENTIALS_PATH as a secret environment variable and paste the complete contents of the downloaded service-account JSON file as its value. This does not require uploading a credentials file to Render.

Do not paste the key into the repository, a committed .env file, or a public support request. If you use a mounted disk on a paid plan, you may instead set GOOGLE_CREDENTIALS_PATH to the credentials file path.

Step 6: Deploy

  1. Go back to your service dashboard
  2. Click "Manual Deploy" → "Deploy latest commit"
  3. Wait for deployment to complete
  4. Check the logs for: 
    ✓ Google Sheets integration enabled for dataset storage

Step 7: Test the Deployment

  1. Get your Render URL from the dashboard (e.g., https://smart-sign-interpreter.onrender.com)

  2. Test the health endpoint:

    curl https://smart-sign-interpreter.onrender.com/api/health

  3. Access the collection tool:

    https://smart-sign-interpreter.onrender.com/collect

  4. Collect a sample gesture and save it

  5. Check that data appears in your Google Sheet

Monitoring and Maintenance

View Logs

In your Render dashboard:

  • Click "Logs" tab
  • Real-time logs appear below
  • Check for errors and Google Sheets connection messages

Update Data Sync

To manually sync data:

# From your local machine
python scripts/sync_google_sheets.py download \
  --credentials /path/to/credentials.json \
  --sheet-id YOUR_SHEET_ID \
  --csv local_data.csv

Retrain Model

The model automatically retrains when you click "RETRAIN" in the data collection tool.

To manually trigger:

curl -X POST https://smart-sign-interpreter.onrender.com/api/model/retrain \
  -H "Content-Type: application/json" \
  -d '{"model_type": "knn"}'

Storage Limits

  • Free tier: Ephemeral disk only (no persistent data on disk)
  • Paid tier: Can add persistent disks
  • Google Sheets: Unlimited (within Google account limits)

Billing

Free tier:

  • 750 hours/month of running services
  • Auto-spins down after 15 minutes of inactivity
  • Good for development/testing

Paid tier:

  • Always-on service
  • More reliable
  • Check Render pricing

Troubleshooting Render Deployment

Service won't start

  1. Check the logs tab

  2. Common issues:

    • Missing environment variables
    • Invalid Google credentials JSON or an incorrect credentials path
    • Port already in use

  3. Click "Manual Deploy" to retry

"Google Sheets integration failed" on Render

  1. Verify GOOGLE_CREDENTIALS_PATH contains the complete valid JSON or matches the credentials file location.
  2. Verify GOOGLE_SPREADSHEET_ID is correct.
  3. Check Google Sheets API is enabled.

Data not appearing in Google Sheets

  1. Check application logs for errors
  2. Verify the spreadsheet is shared with the service account
  3. Try accessing the collection tool and checking /api/dataset/stats

Service goes to sleep

On Render's free tier, services auto-spin down after 15 minutes of inactivity.

  • This is normal
  • Your Google Sheets data is safe
  • Data restores when you make a request
  • Upgrade to paid tier for always-on service

Advanced Configuration

Custom Domain

  1. In your service settings, go to "Custom Domain"
  2. Enter your domain (e.g., gesture.example.com)
  3. Update your domain's DNS records as instructed

Scaling

If you need better performance:

  1. Upgrade to a paid plan
  2. Consider using Render's PostgreSQL for session storage
  3. Cache model predictions if needed

CI/CD

Render automatically deploys on push to main branch. To disable:

  1. Go to service settings
  2. Turn off "Auto-deploy"

Rendering Service vs Local Development

Aspect Local Render
Data persistence Disk based Google Sheets
Uptime Only while running 24/7 (paid) or auto-sleep (free)
Performance Fast local access Network latency to Google
Cost Free Free/Paid
Model updates Manual via CLI Via web interface

Next Steps

  1. Set up real glove hardware to send data
  2. Collect training data over time
  3. Monitor model accuracy
  4. Share deployment URL with team
  5. Consider upgrading to paid plan for production use

Support

Maintenance Checklist

  • Google Sheets API enabled
  • Service account credentials downloaded
  • Google Sheet created and shared
  • Environment variables configured
  • Credentials stored as a Render secret (or uploaded to mounted storage)
  • Service deployed successfully
  • Test data collection working
  • Google Sheets receiving data
  • Model retraining works
  • Share deployment URL with team