PodKiller

A chaos engineering dashboard for Kubernetes that provides a TIE Fighter-themed interface for killing pods and monitoring cluster resilience. Built with Express.js backend and vanilla JavaScript frontend.

Features

Core Functionality

• Pod Termination: Kill specific pods or random pods within selected namespaces
• Minikube Management: Start and stop Minikube clusters directly from the interface
• Namespace Selection: Browse and select from available Kubernetes namespaces
• Real-time Monitoring: Watch pod status changes and recovery times
• Latency Injection: Add network latency to selected pods for chaos testing

Dashboard Components

• Activity Log: Real-time logging of all operations and status changes
• Pod Grid: Visual representation of pods with status indicators (Running, Pending, Terminating)
• Recovery Analytics: Chart showing pod recovery times over time
• Session Statistics: Track pods killed, average recovery time, and failed recoveries

Demo App

• Chaos Clock: Counter app that shuts down when a pod is killed - refer to demo-app folder for more details

Prerequisites

• Node.js (v14 or higher)
• Minikube installed and configured
• kubectl configured to work with your Minikube cluster

Installation

1. Fork and Clone the repository:

git clone <https://github.com/oslabs-beta/podkiller>
cd podkiller

2. Install dependencies:

npm install

3. Start the server:

npm start

4. Open your browser to http://localhost:3000

Usage

Getting Started

1. Start Minikube: Click the connection status button to start Minikube if not already running
2. Select Namespace: Choose a namespace from the dropdown menu
3. View Pods: Pods in the selected namespace will appear in the grid

Killing Pods

• Random Kill: Click "Kill Pod" without selecting any pods to kill a random pod
• Specific Kill: Select one or more pods by clicking on them, then click "Kill Pod"
• Select All: Use "Select All Pods" to select/deselect all visible pods

Adding Latency

1. Select one or more pods
2. Set the desired latency in milliseconds (default: 500ms)
3. Click "Add Lag" to inject network latency
4. As the app continues making requests through the delayed network, the average keeps climbing
5. Eventually it may timeout completely and go offline

Monitoring

• Activity Log: View real-time logs of all operations
• Recovery Chart: Monitor pod recovery times over multiple chaos events
• Statistics Panel: Track session metrics and reset counters

API Endpoints

Cluster Management

• GET /api/status - Check Minikube connection status
• POST /api/minikube/start - Start Minikube cluster
• POST /api/minikube/stop - Stop Minikube cluster

Kubernetes Operations

• GET /api/namespaces - List available namespaces
• GET /api/pods?namespace=<name> - List pods in specified namespace
• GET /api/killpod?namespace=<name>&podNames=<array> - Kill specified pods (Server-Sent Events)

Chaos Engineering

• POST /api/latency/set - Inject latency into pod
• POST /api/latency/clear - Remove latency from pod

Reporting

• GET /api/reports - Retrieve chaos engineering reports
• POST /api/reports - Save new chaos engineering report

Key Components

Frontend Modules

• app.js: Main application state, UI interactions, and API communication
• chart.js: Chart.js integration for recovery time visualization and session statistics
• canvas.js: Animated starfield background and laser beam effects

Backend Modules

• server.js: Express server with Kubernetes API integration
• killpod.js: Core pod termination logic with recovery monitoring
• latency.js: Network latency injection functionality

Configuration

The application uses Minikube's default configuration. Ensure your kubectl context is set to the Minikube cluster you want to manage.

Safety Features

• Connection status monitoring prevents operations when cluster is unavailable
• Operation locks prevent concurrent Minikube start/stop operations
• Automatic cleanup of terminated pods and failed operations
• Comprehensive error handling and user feedback

Monitoring and Analytics

The dashboard automatically tracks:

• Recovery times for each killed pod
• Session statistics (pods killed, average recovery time, failures)
• Historical data stored in JSON reports
• Real-time pod status updates via Server-Sent Events

Troubleshooting

Common Issues

1. Minikube won't start: Ensure Minikube is properly installed and has sufficient resources
2. No pods visible: Check that the selected namespace contains running pods
3. Connection issues: Verify kubectl can communicate with your cluster
4. Permission errors: Ensure proper RBAC permissions for pod deletion

Logs

Check the browser console and server logs for detailed error information. The activity log in the dashboard shows real-time operation status.

Contributing

This is a chaos engineering tool designed for testing and learning. Use responsibly in development and testing environments only.

Warning

This tool is designed for chaos engineering and testing purposes. Do not use in production environments without proper safeguards and understanding of the potential impact.