Couldn't pass 200 request / second

[rendybjunior]

Description of issue

I tried to increase concurrent user to 100,000 and decrease min wait time to 1ms, and max wait time 2ms.
I couldn't pass 200 request / second, after increasing concurrent user and decreasing wait time.

Expected behavior

I could achive hundred thousand request / second

Actual behavior

I couldn't pass 200 request / second

Environment settings (for bug reports)

• OS: Ubuntu 14.04
• Python version: 2.7
• Locust version: 0.8.1

[heyman]

What happens when you do this? Do you get errors? Have you increased the OS limit on the max number of open files? How much CPU does the locust/python process use?

[cgoldberg]

I could achieve hundred thousand request / second

without knowing specifics of your environment and tests, it's impossible to say if this is feasible. It depends on your machine specs, network topology, configuration, number of slaves running, and many other factors. You need to monitor all resources to figure out where your bottleneck is.

btw, I hope you are running in distributed mode with many slave machines. You'll never get even close to 100,000 concurrent users from a single load generator.

[aldenpeterson-wf]

I don't really see anything actionable here. Any expectation that a single Locust node can produce hundreds of thousands of requests per second really isn't realistic.