Locust — Scalable Load Testing in Pure Python

What it is

Locust is an open-source load testing tool where you define user behavior in plain Python code. There is no DSL to learn, no XML configuration, and no GUI-driven test creation. Locust is distributed by design: run workers across multiple machines to generate massive load. A real-time web UI shows request statistics, response times, and failure rates as tests run.

Locust targets Python developers and QA teams who want to write performance tests using familiar Python tools and libraries. It suits API load testing, website performance testing, and any HTTP-based workload.

How it saves time or tokens

Locust tests are Python classes: define tasks as methods, use Python libraries for data generation, and leverage pytest fixtures or factories for test data. The Python ecosystem (requests, faker, databases) is available in your test scripts. No proprietary format means tests are readable, maintainable, and version-controllable.

For AI-assisted development, Locust's Python-based approach means LLMs generate test scripts using familiar Python patterns.

How to use

1. Install Locust: pip install locust.
2. Create a locustfile.py defining user classes and tasks.
3. Run: locust and open the web UI at http://localhost:8089.

Example

from locust import HttpUser, task, between

class WebsiteUser(HttpUser):
    wait_time = between(1, 3)

    @task(3)
    def view_homepage(self):
        self.client.get('/')

    @task(1)
    def view_product(self):
        self.client.get('/api/products/1')

    @task(1)
    def create_order(self):
        self.client.post('/api/orders', json={
            'product_id': 1,
            'quantity': 2,
        })

    def on_start(self):
        self.client.post('/api/login', json={
            'username': 'testuser',
            'password': 'testpass',
        })

Related on TokRepo

• AI tools for testing -- testing and QA tools
• AI coding tools -- Python development tools

Common pitfalls

• Running all workers on a single machine. Locust workers are CPU-bound; distribute them across machines to generate real load without bottlenecking on the test runner.
• Not using wait_time between tasks. Without delays, Locust hammers the server unrealistically fast, producing results that do not reflect real user behavior.
• Ignoring the web UI statistics. The real-time dashboard shows percentile response times, failure rates, and throughput that help identify performance regressions during the test.