Locust introduces installation and basic usage

1. Introduction to Locust

Locust (locust) is an open source performance testing tool, the main idea is to simulate a group of users accessing your system

2. Features

(1) Define user behavior in code

No need to install bulky software, just simple Python code

(2) Distributed and scalable

Locust supports running load tests on multiple machines, so it can be used to simulate requests from millions of users

(3) Proven and battle tested

Locust is used in many real projects

(4) Locust has a neat HTML+JS user interface that displays relevant test details in real time

Since the user interface is web-based, it is cross-platform and easily extensible

2. Locust installation

Install command:

pip install locust

Successful installation verification:

locust --version    # 查看安装的版本

locust -h           # 查看使用帮助

Note: The version of python is preferably above 3.0

3. How to use Locust

1. Core steps

Define task (interface request): ordinary function, must have a formal parameter, must use @task to modify the function
Define task set (user behavior): a class must inherit TaskSet, and the defined tasks are inside the class
Define user class (user): a class that must inherit from HttpUser, configure the host address, override tasks, the format is a list or dictionary, and the value is the name of the defined task set
Start the service, configure the execution: locust -f file name

example:

from locust import HttpUser, TaskSet, task


# 定义任务集
class TaskTest(TaskSet):
    # 定义任务
    @task
    def say(params):
        print("正在说话")

    @task
    def sing(params):
        print("正在唱歌")


# 定义用户类
class Test(HttpUser):
    tasks = [TaskTest]
    host = "http://localhost"

Execute the locust step:

(1) Open the pycharm terminal and enter locust -f file name to run

(2) Open the browser and enter localhost:8089 to run the locust control interface

(3) Enter data and click start swarming to execute the code

Number of users (peak concurrency): the number of concurrent users is the number of users
Spawn rate (users started/second): The number of users loaded per second
Host (eg http://www.example.com): host address, if there is one in the code, it can be executed by default

2. Commonly used methods for task sets

(1) on_start: pre-method (pre-task), called once before all tasks

(2) on_stop: post method (post task), called once when the task set stops

(3) @task(weight): Modifies the function and defines the function as a task. weight represents the weight, the greater the weight, the more execution times, the default value is 1

from locust import TaskSet, task


# 定义任务集
class TaskTest(TaskSet):
    # 定义任务
    @task(10)
    def say(params):
        print("正在说话")

    @task(5)
    def sing(params):
        print("正在唱歌")

    def on_start(self):
        self.say()

    def on_stop(self):
        self.sing()

3. Common attributes of user classes

(1) min_wait: The lower bound of the waiting time between user execution tasks, unit: milliseconds, default value: 1000

(2) max_wait: The upper bound of the waiting time between user execution tasks, unit: milliseconds, default value: 1000

(3) host: URL of the application under test, for example: http://localhost

(4) weight: the probability of the user being selected, the greater the weight, the greater the chance of being selected, the default value: 10

from locust import HttpUser

# 定义用户类
class Test(HttpUser):
    tasks = [TaskTest]
    host = "http://localhost"
    min_wait = 1000
    max_wait = 2000
    weight = 10

4. Locust distributed application - web interface implementation

(1) Host (control machine)

locust -f 目录\执行文件名 --master

(2) Slave host (executive machine)

locust -f 目录\执行文件名 --worker --master-host=主机ip地址 --master-port=主机端口

Key points that slave hosts must rely on:

Must have python and locust environment

Must have a copy of the host script

If the slave host is on the same machine as the master, --master-host and --master-port can be omitted

5. Locust distributed application - implementation of no-graph mode (non-web interface)

locust -f 目录\执行文件名 --master --headless -u 200 -r 20 --expect-workers 2 --run-time 20s --csv ./result.csv


--headless:不使用web界面
-u:虚拟用户数
-r:每秒生成的用户数
--expect-workers:执行机数量
--run-time:运行时间（h:小时,m:分钟,s:秒）
--csv:保存执行结果

4. Application of Locust in interface testing

Simple case implementation:

from locust import HttpUser, TaskSet, task

headers = {"content-type": "application/json;charset=UTF-8"}
data = {"username": "hogwarts", "passwords": "test12345", "code": ""}
params = {"page": 1, "limit": 20, "goodsSn": "11111", "name": ""}


# 定义任务集
class TaskTest(TaskSet):

    # 定义任务
    # 登录方法
    @task(1)
    def login(r):
        response = r.client.post(url="/admin/auth/login", json=data, headers=headers)
        headers["x-litemall-admin-token"] = response.json()["data"]["token"]
        print(response.json())

    # 获取商品列表方法
    @task(1)
    def list(r):
        response = r.client.get(url="/admin/goods/list", params=params, headers=headers)
        print(response.json())

    def on_start(self):
        self.login()

    def on_stop(self):
        self.list()

# 定义用户类
class Test(HttpUser):
    tasks = [TaskTest]
    host = "https://litemall.hogwarts.ceshiren.com"
    min_wait = 1000
    max_wait = 3000

operation result: