kong easily gets overwhelmed when receiving many requests, resulting in an error complaining about not having enough workers. This was due to KONG_NGINX_WORKER_PROCESSES=1 being set. Kong automatically determins the right number of worker processes when this is not specified, which resolves the issue.

What kind of change does this PR introduce?

Bug fix (removing a bad env var for the Kong container).

What is the current behavior?

If many parallel requests are made to the Storage API, Kong will quickly stop responding and start terminating socket connections.

What is the new behavior?

Kong is able to handle many parallel connections without choking.