Pro-tip: Scaling properly with Nginx worker_process and max_pool_size

This post talks about worker_process and passenger_max_pool_size options for nginx and how these configurations play an important role for scalability.

Recently our client reported that API performance was  bad and it was taking a few minutes to respond! But, when we tested this API this was quite efficient and was responding in few milliseconds. What was wrong? As the number of users increased, the number of concurrent requests to the server had also increased. Initially, it was only 100’s of request per second coming to a server but now it was about 4 k requests per second and we realised that we had not scaled our server to handle this load.

When this problem arrived I thought of breaking down the problem in steps.

• First I checked RDS instance. It was using 20%
• I revisited all queries they were optimized enough. So database was not bottleneck
• Sent unauthorized request to the server so with less processing…