Sandbox Testing

Define health checks for all services to ensure they’re fully ready:

1
healthcheck:
2
  test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
3
  interval: 10s
4
  timeout: 5s
5
  retries: 5
6
  start_period: 30s

Health checks prevent Pensar from starting tests before services are ready.

Specify service dependencies with conditions:

1
depends_on:
2
  db:
3
    condition: service_healthy
4
  cache:
5
    condition: service_started

This ensures proper startup order and prevents race conditions.

Include database seeding in your setup:

1
db:
2
  image: postgres:15
3
  volumes:
4
    - ./seed-data.sql:/docker-entrypoint-initdb.d/seed-data.sql

Test data helps the AI agent understand your data model and test more effectively.

Expose all ports that Pensar needs to test:

1
ports:
2
  - "3000:3000"  # Main application
3
  - "8080:8080"  # Admin panel
4
  - "4000:4000"  # GraphQL endpoint

This ensures all endpoints are accessible for testing.

Your application can be started with:

• A Docker Compose file
• A simple shell script
• Standard development commands (npm start, rails server, etc.)

If your entire dev environment can run locally with minimal configuration, sandbox testing is perfect.

You want to safely test exploits that:

• Modify or delete data
• Upload malicious files
• Execute system commands
• Cause service disruptions

Sandboxes allow aggressive testing without production risk.

You’re using auto-remediation and want to:

• Replay POCs to verify fixes work
• Iterate on patches until confirmed resolved
• Have high confidence in automated fixes

Sandbox environments enable real-time verification loops.

You need:

• Identical environments for every scan
• No environmental drift
• Reproducible results
• Fresh state for each test run

If your sandbox environment fails to start:

1. Check the sandbox logs in the Pensar Console scan details
2. Verify your Docker Compose file works locally
3. Ensure health checks are properly configured
4. Check for missing environment variables
5. Review build logs for dependency installation errors

The AI agent automatically attempts to fix configuration issues, but complex problems may require manual intervention.

If services fail health checks:

• Increase the start_period in health check configuration
• Verify health check commands are correct
• Check service logs for startup errors
• Ensure dependencies are available (network, volumes)

If tests fail to reach your application:

• Verify ports are correctly exposed in Docker Compose
• Check that the application is listening on 0.0.0.0, not localhost
• Review application logs for startup errors
• Ensure the application waits for dependencies before starting

If sandboxes take too long to start:

• Optimize Docker image sizes
• Use build caching effectively
• Pre-build images and push to a registry
• Reduce the number of dependencies
• Consider switching to live environment testing for very complex setups

Minimize sandbox startup time:

• Use lightweight base images
• Leverage Docker layer caching
• Pre-install dependencies in container images
• Use multi-stage builds
• Avoid unnecessary services in dev environment

Ensure all services have proper health checks: - Database connections are ready - Application server is accepting requests - Migrations have completed - External dependencies are available

Provide test data that represents your production data: - User accounts with different roles - Sample products, orders, content - Edge cases and boundary conditions - Data that exercises your business logic

Track sandbox metrics over time:

• Average startup time
• Success/failure rates
• Resource utilization
• Testing duration

Use this data to optimize your configuration.