Stress Testing

💡 Simulating stress test of 5000 Users

Introduction

Load testing evaluates application performance under expected user loads. This guide outlines load testing with 5000 users using Locust, including multiple endpoints and graph generation.

Prerequisites:

1. Python 3.8 or later.
2. Locust. Follow Guide here

Load Testing Setup:

1. Create load_test.py.
2. Define user behavior with HttpUser class and tasks.
3. Set user count and hatch rate in Locust web interface.

Script:

This script here is for reference. Actual script will be shared to testing team.

from locust import HttpUser, task, between
import pandas as pd
import matplotlib.pyplot as plt

response_times = []

class MyUser(HttpUser):
    wait_time = between(1, 5)
    
    @task
    def hit_endpoint1(self):
        response = self.client.get("/endpoint1")
        response_times.append(response.elapsed.total_seconds() * 1000)
        
    @task
    def hit_endpoint2(self):
        response = self.client.get("/endpoint2")
        response_times.append(response.elapsed.total_seconds() * 1000)
        
    # Add more @task methods for endpoints

# Access web interface: http://localhost:8089

# Save response times to a CSV file
data = pd.DataFrame(response_times, columns=['ResponseTime'])
data.to_csv('response_times.csv', index=False)

Running the Load Test:

1. Open terminal/command prompt.
2. Navigate to script location: cd path/to/script.
3. Run Locust: locust -f load_test.py.
4. Configure user count, hatch rate, and start the test.

Access web interface:

💡Click View Graph to view the real-time graph, or access the graph here: http://localhost:8089

Graph Generation and Reporting:

1. After load test, open terminal.
2. Run graph script: python graph.py.

# graph.py
import pandas as pd
import matplotlib.pyplot as plt
import uuid

# Load response times from the CSV file
data = pd.read_csv('response_times.csv')
response_times = data['ResponseTime'].tolist()

# Generate a graph using matplotlib and save it with a random file name
file_name = f'response_time_histogram_{uuid.uuid4()}.png'
plt.figure(figsize=(10, 6))
plt.hist(response_times, bins=20, color='blue', alpha=0.7)
plt.xlabel('Response Time (ms)')
plt.ylabel('Frequency')
plt.title('Distribution of Response Times')
plt.grid(True)
plt.savefig(file_name)  # Save the graph with the random file name

Access the graph png that will be saved to local machine.

Conclusion:

Locust enables in-depth load testing with multiple endpoints. Monitor real-time metrics and generate custom graphs for performance optimization insights.