Search code examples
locust

Strange locust behavior in extreme cases

I am working on a Ubuntu desktop and try to run load testing to check our server sites capacity. Usually my load testing is of 500 to 2000 users with the time range of 4 to 12 hours. And everything runs smoothly before it ends on time.

Today, I was trying to see how the server site behaves, in some extreme case. So I have it executed with 10000 users for testing in 10 minutes, ramp-up rate is 1000.

Here is my running command: locust -f locustTesting.py --host=http://server.site -u 10000 -t 10m -r 1000 --html testing.html .
a) locustTesting.py is my test script, that contains a list of subsequent API calls.
b) testing.html would be the test results.

As of now, I have seen such diagram from http://127.0.0.1:8089 site.

image

As you may find:

  1. it is 80 minutes passed the starting time, and the locust script is still on-going !
  2. 10 minutes later, the number has dropped from 10000 to about 100, on this bottom user chart.

So my questions are:

  1. What happen 10 minutes after ? Are they suppose to be stop by then ?
  2. If I did receive my "testing.html" in the end, is that the data contains the full-run, or just the first 10 minutes?
  3. In order to get the this testing.html file, I have to keep waiting, right ?

Thanks for the tips.

Jack

Solution

  • I was using locust of version 2.20.1. Thanks to Cyberwiz's suggestion, I have upgraded to 2.30.1, and I no longer see this issue.

    problem solved.