Your first test

A Locust test is essentially just a Python program making requests to the system you want to test. This makes it very flexible and particularly good at implementing complex user flows. But it can do simple tests as well, so let’s start with that:

from locust import HttpUser, task

class HelloWorldUser(HttpUser):
    @task
    def hello_world(self):
        self.client.get("/hello")
        self.client.get("/world")

This user will make an HTTP request to /hello, then to /world, and then repeat. For a full explanation and a more realistic example see Writing a locustfile.

Change /hello and /world to some actual paths on the website/service you want to test, put the code in a file named locustfile.py in your current directory and then run locust:

$ locust
[2021-07-24 09:58:46,215] .../INFO/locust.main: Starting web interface at http://0.0.0.0:8089
[2021-07-24 09:58:46,285] .../INFO/locust.main: Starting Locust 2.31.9.dev42

Locust’s web interface

Open http://localhost:8089

_images/webui-splash-screenshot.png
Provide the host name of your server and try it out!

The following screenshots show what it might look like when running this test using 50 concurrent users, with a ramp up rate of 1 user/s

_images/webui-running-statistics.png
Under the Charts tab you’ll find things like requests per second (RPS), response times and number of running users:
_images/bottlenecked_server.png

Note

Interpreting performance test results is quite complex (and mostly out of scope for this manual), but if your graphs start looking like this, the target service/system cannot handle the load and you have found a bottleneck.

When we get to around 20 users, response times start increasing so fast that even though Locust is still spawning more users, the number of requests per second is no longer increasing. The target service is “overloaded” or “saturated”.

If your response times are not increasing then add even more users until you find the service’s breaking point, or celebrate that your service is already performant enough for your expected load.

If you’re having trouble generating enough load to saturate your system, take a look at Increasing the request rate.

Direct command line usage / headless

Using the Locust web UI is entirely optional. You can supply the load parameters on the command line and get reports on the results in text form:

$ locust --headless --users 10 --spawn-rate 1 -H http://your-server.com
[2021-07-24 10:41:10,947] .../INFO/locust.main: No run time limit set, use CTRL+C to interrupt.
[2021-07-24 10:41:10,947] .../INFO/locust.main: Starting Locust 2.31.9.dev42
[2021-07-24 10:41:10,949] .../INFO/locust.runners: Ramping to 10 users using a 1.00 spawn rate
Name              # reqs      # fails  |     Avg     Min     Max  Median  |   req/s failures/s
----------------------------------------------------------------------------------------------
GET /hello             1     0(0.00%)  |     115     115     115     115  |    0.00    0.00
GET /world             1     0(0.00%)  |     119     119     119     119  |    0.00    0.00
----------------------------------------------------------------------------------------------
Aggregated             2     0(0.00%)  |     117     115     119     117  |    0.00    0.00
(...)
[2021-07-24 10:44:42,484] .../INFO/locust.runners: All users spawned: {"HelloWorldUser": 10} (10 total users)
(...)

See Running without the web UI for more details.

More options

To run Locust distributed across multiple Python processes or machines, you start a single Locust master process with the --master command line parameter, and then any number of Locust worker processes using the --worker command line parameter. See Distributed load generation for more info.

To see all available options type: locust --help or check Configuration.

Next steps

Now, let’s have a more in-depth look at locustfiles and what they can do: Writing a locustfile.