LoadForge tests are designed to be used as semi-static definitions for repeatable tests against your service. You specify the site to test and the load to generate and can then trigger a run at any time.
Browse to the Tests menu item and click "Create new test" to make your first test. You will be asked to complete a form with these key variables:
Name of test
A display purposes only name, to help you identify the test. Common name formats would be descriptive, such as "MyWebsite Submission API", or "Staging Server".
Servers to launch
The number of droplets (DigitalOcean cloud instances) we will create as a part of your test. They are created and then destroyed automatically. Generally speaking a single droplet will generate around 10,000 users. We always launch a controller, and then however many droplets you specify.
Users to simulate
The number of users to simulate across the entire test. They will be divided amongst all the worker nodes (not the controller) at launch. The load an individual user generates is governed largely by the "wait_time" field in your locustfile. Our default is:
wait_time = between(5, 9)
This means wait 5 to 9 seconds per page as a fake user pretending to consume the content of the page. This is a good general rule.
Spawn rate of users
How quickly to scale up from 0 users to the "Users to simulate". Scaling too quickly can cause errors and be an unrealistic burst of traffic on your webserver. We recommend around 2-5% of your total user amount.
One of your validated hosts to run the test against. This ultimately becomes something like https://loadforge.com:443/
Where in the world to run the tests from. This is good for testing latency, deploying tests closest to your users, etc. We support:
New York, San Francisco, London, Amsterdam, Singapore, France, Toronto, Bangalore
For advanced users we allow customization of the actual test script. The default will simply request the index page of your website. For more details on this see the locustfile section of the docs.