I am learning to design scalable system, for now using Windows machine. I created two servers that will listen to port 27016 and 27015, all they do is return "HelloWorld!" response. I had set listen(ListenSocket, SOMAXCONN) for both the servers when creating them in Visual studio following Winsock tutorial. Using jmter performed load test on each of them individually (1000 request per sec) and got everything OK.
Now when I introduced nginx which is listening to port 80 and load balancing the requests (1000 req per sec) among the two servers I mentioned above, many requests are being dropped down while performing load test using jmeter.
I am assuming that queue size for port 80 is not configured for high traffic and want to tune it. How to set the queue size to maximum possible value either from nginx config or cmd command?
Best Answer
Here is the information from Nginx web site:
So, the best choice for worker_processes is the same as the number of cores you have. And the value or worker_connections defaults to 512. You can increase it depending on your needs, and as your system can handle.
The following could be helpfull too:
I hope this will helps!
Note: I agree with @Alexey Ten