LoadPilot

Automated Load Testing Platform for Kubernetes + JMeter + AI

LoadPilot helps you find the optimal pod configuration for your Kubernetes workloads by testing different CPU/memory/replica combinations. It supports JMeter test plans and uses AI (Ollama) to provide intelligent recommendations.

---

Features

Feature	Description
Matrix Testing	Test all CPU/memory/replica combinations to find optimal K8s config
JMX Testing	Execute existing JMeter test plans (.jmx files)
AI Testing	Interactive tests with real-time AI-powered recommendations
Test Controls	Start, stop, pause, resume, scale load dynamically
Cost Analysis	Real-time cloud cost calculation (AWS, GCP, Azure)
Performance Scoring	Weighted scoring based on latency, throughput, cost, errors
Kubernetes Integration	Connect to real clusters, deploy test pods, scale dynamically

---

Prerequisites

Component	Required	Install
Python	Yes	3.9+
Node.js	No	nodejs.org
JMeter	No	jmeter.apache.org
Ollama	No	ollama.ai

---

Quick Start

1. Install Dependencies

cd loadpilot/backend
pip install -r requirements.txt

2. Start Server

cd loadpilot
./start.sh

Server runs at: http://localhost:8000

3. Open API Docs

Visit http://localhost:8000/docs for interactive API documentation.

---

Usage Examples

Example 1: Find Optimal Kubernetes Config (Matrix Testing)

# Create a test that tries all combinations of CPU/RAM/replicas
curl -X POST http://localhost:8000/api/tests \
  -H "Content-Type: application/json" \
  -d '{
    "name": "api-gateway",
    "cloud_provider": "aws",
    "region": "us-east-1",
    "resource_config": {
      "cpu_min": 250,
      "cpu_max": 2000,
      "cpu_step": 250,
      "memory_min": 256,
      "memory_max": 4096,
      "memory_step": 512
    },
    "replica_config": {
      "min": 1,
      "max": 5
    }
  }'

# Run the test
curl -X POST http://localhost:8000/api/tests/<TEST_ID>/run

# Get results
curl http://localhost:8000/api/tests/<TEST_ID>

Example 2: Run JMeter Test Plan (JMX Testing)

# Upload and run a JMeter .jmx file
curl -X POST http://localhost:8000/api/tests/jmx \
  -F "name=api-load-test" \
  -F "cloud_provider=aws" \
  -F "threads=100" \
  -F "ramp_up=30" \
  -F "duration=120" \
  -F "jmx_file=@/path/to/test.jmx"

# Run the test
curl -X POST http://localhost:8000/api/tests/<TEST_ID>/run-jmx

# Get results
curl http://localhost:8000/api/tests/<TEST_ID>

Example 3: Interactive Test with AI (Ollama)

# Start Ollama (in separate terminal)
ollama serve
ollama pull llama3.2

# Check if Ollama is available
curl http://localhost:8000/api/ai/check

# Start an interactive load test
curl -X POST http://localhost:8000/api/load-test/start \
  -F "name=interactive-test" \
  -F "threads=50" \
  -F "duration=300" \
  -F "jmx_file=@/path/to/test.jmx"

# Monitor status
curl http://localhost:8000/api/load-test/<TEST_ID>/status

# Increase load by 25%
curl -X POST http://localhost:8000/api/load-test/<TEST_ID>/increase

# Decrease load by 50%
curl -X POST http://localhost:8000/api/load-test/<TEST_ID>/decrease?percent=50

# Scale to specific threads
curl -X POST "http://localhost:8000/api/load-test/<TEST_ID>/scale?threads=200"

# Get AI recommendation based on current results
curl -X POST http://localhost:8000/api/load-test/<TEST_ID>/ai-decide

# Pause the test
curl -X POST http://localhost:8000/api/load-test/<TEST_ID>/pause

# Resume the test
curl -X POST http://localhost:8000/api/load-test/<TEST_ID>/resume

# Stop the test
curl -X POST http://localhost:8000/api/load-test/<TEST_ID>/stop

---

API Reference

Core Endpoints

Method	Endpoint	Description
GET	/api/dashboard	Dashboard statistics
POST	/api/tests	Create matrix test
POST	/api/tests/jmx	Create JMX test
GET	/api/tests	List all tests
GET	/api/tests/:id	Get test details
POST	/api/tests/:id/run	Run matrix test
POST	/api/tests/:id/run-jmx	Run JMX test
DELETE	/api/tests/:id	Delete test

Interactive Test Controls

Method	Endpoint	Description
POST	/api/load-test/start	Start interactive test
GET	/api/load-test/:id/status	Get test status
POST	/api/load-test/:id/stop	Stop test
POST	/api/load-test/:id/pause	Pause test
POST	/api/load-test/:id/resume	Resume test
POST	/api/load-test/:id/scale	Scale to specific threads
POST	/api/load-test/:id/increase	Increase load (%)
POST	/api/load-test/:id/decrease	Decrease load (%)

AI Endpoints

Method	Endpoint	Description
GET	/api/ai/check	Check Ollama status
POST	/api/ai/recommend/:id	Get AI recommendations
POST	/api/ai/summary/:id	Get AI summary
POST	/api/load-test/:id/ai-decide	AI decision during test

Utility Endpoints

Method	Endpoint	Description
GET	/api/jmeter/check	Check JMeter installation

Kubernetes Endpoints

Method	Endpoint	Description
GET	/api/k8s/status	Get cluster connection status
POST	/api/k8s/connect	Connect to Kubernetes cluster
POST	/api/k8s/disconnect	Disconnect from cluster
GET	/api/k8s/namespaces	List namespaces
POST	/api/k8s/deploy	Deploy test pods
GET	/api/k8s/deployments	List LoadPilot deployments
GET	/api/k8s/deployments/:id	Get deployment details
DELETE	/api/k8s/deployments/:id	Delete deployment
POST	/api/k8s/deployments/:id/scale	Scale deployment
GET	/api/k8s/metrics	Get cluster metrics

---

Parameters

Matrix Test Parameters

Resource Config

Parameter	Type	Default	Description
cpu_min	int	250	Min CPU (millicores)
cpu_max	int	2000	Max CPU (millicores)
cpu_step	int	250	CPU increment
memory_min	int	256	Min memory (MB)
memory_max	int	4096	Max memory (MB)
memory_step	int	256	Memory increment

Replica Config

Parameter	Type	Default	Description
min	int	1	Min replicas
max	int	5	Max replicas

JMX Test Parameters

Parameter	Type	Required	Description
name	string	Yes	Test name
cloud_provider	string	No	aws/gcp/azure
region	string	No	Cloud region
threads	int	No	Virtual users (default: 100)
ramp_up	int	No	Ramp-up time in sec (default: 60)
duration	int	No	Test duration in sec (default: 300)
jmx_file	file	Yes	JMeter .jmx file

---

Scoring System

Each configuration is scored 0-10:

score = (latency × 0.30) + (throughput × 0.30) + (cost × 0.20) + (errors × 0.20)

Factor	Weight	Better =
Latency (P99)	30%	Lower
Throughput (RPS)	30%	Higher
Cost ($/mo)	20%	Lower
Error Rate	20%	Lower

---

Architecture

┌─────────────────────────────────────┐
│              Frontend               │
│        React + Vite + Recharts     │
└──────────────────┬────────────────┘
                   │ HTTP/REST
┌──────────────────▼────────────────┐
│              Backend               │
│         FastAPI + SQLite          │
├───────────────────────────────────┤
│  Matrix    │ Cost     │ JMX      │
│  Simulator │ Calculator│ Runner   │
│  AI Advisor│ Ollama   │ Controls  │
└───────────────────────────────────┘

---

Project Structure

loadpilot/
├── README.md
├── SPEC.md
├── requirements.txt
├── setup.sh
├── start.sh
├── stop.sh
├── backend/
│   ├── main.py              # FastAPI application
│   ├── models.py            # Pydantic models
│   ├── simulator.py          # Matrix test simulator
│   ├── cost_calculator.py    # Cloud cost calculation
│   ├── ollama_client.py      # AI/Ollama integration
│   ├── test_runner.py        # Interactive test runner
│   └── k8s_client.py        # Kubernetes client
└── frontend/
    ├── package.json
    ├── vite.config.js
    └── src/
        ├── App.jsx
        ├── index.css
        └── pages/
            └── Dashboard.jsx

---

Installing Optional Components

JMeter

# macOS
brew install jmeter

# Ubuntu/Debian
sudo apt install jmeter

# Manual
# Download from https://jmeter.apache.org/

Ollama (for AI features)

# macOS
brew install ollama

# Linux
curl -fsSL https://ollama.ai/install.sh | sh

# Start Ollama
ollama serve

# Pull a model
ollama pull llama3.2

---

Example 4: Kubernetes Integration

# Connect to Kubernetes cluster (uses kubeconfig or in-cluster config)
curl -X POST "http://localhost:8000/api/k8s/connect?namespace=default"

# Check connection status
curl http://localhost:8000/api/k8s/status

# Deploy test pods with specific configuration
curl -X POST "http://localhost:8000/api/k8s/deploy?name=load-test&cpu_millicores=500&memory_mb=512&replicas=2&image=nginx&namespace=default"

# List all LoadPilot deployments
curl http://localhost:8000/api/k8s/deployments

# Scale a deployment (replace DEPLOYMENT_ID)
curl -X POST "http://localhost:8000/api/k8s/deployments/DEPLOYMENT_ID/scale?replicas=4"

# Get cluster metrics
curl "http://localhost:8000/api/k8s/metrics?namespace=default"

# Delete a deployment
curl -X DELETE "http://localhost:8000/api/k8s/deployments/DEPLOYMENT_ID"

# Disconnect from cluster
curl -X POST http://localhost:8000/api/k8s/disconnect

---

Prerequisites (Kubernetes)

Component	Required	Install
kubectl	Yes	kubernetes.io/docs
kubeconfig	Yes	~/.kube/config or cluster config

Setup kubectl

# macOS
brew install kubectl

# Verify installation
kubectl version --client

Configure Cluster Access

# Default kubeconfig (~/.kube/config)
kubectl config get-contexts

# For specific cluster
kubectl config use-context my-cluster

---

Roadmap

• Matrix testing (CPU/RAM/Replicas)
• JMX file execution
• AI recommendations (Ollama)
• Test controls (stop/start/scale)
• Real Kubernetes integration
• CI/CD pipeline hooks
• GitOps output (Helm/Kustomize)
• Historical trend analysis
• Slack/Teams notifications

---

License

MIT