High Volume Testing, Identifying Break Points

Now it’s time to generate a high volume of traffic to identify the NGINX Proxy’s breaking points.

  1. Run the Locust test with a high volume of Users

Number of Users: 2000

Spawn rate: 200

Host: http://10.1.1.9/

Advanced Options, Run time: 30s

../../_images/locus-2000-200-30.png

Review the Locust result charts

Note

Did you notice any changes to the graphs? Were there any failures?

If you’re seeing failures, review the Failures tab at the top of the Locust web interface.

  1. Review the logs on NGINX Proxy’s command line interface for failure reasons
sudo grep crit /var/log/nginx/error.log
Copy to clipboard

Note

What problem is identified in the NGINX Proxy’s error logs?

3. Fix the problem by increasing the rlimit This changes the limit on the number of open files that a worker process may have

In NGINX Instance Manager, edit the nginx.conf file.

Increase rlimit to 4096, by uncommenting line 4

  • worker_rlimit_nofile 4096;

Increase worker_connections from 1024 to 4096

  • worker_connections 4096; (line 11 or 12)

Publish the new configuration.

../../_images/nim-rlimit-4096.png
  1. Run the same test again

Review the resulting Locust graphs

Note

Were there improvements and were there still failures?