Topic: (10601)WSAECONNREFUSED error while connecting to Apache

[prog_indigo]

There is windows server 2012, it has a web application installed on apache + mod_fcgid + php-cgi + postgresql.
The application is intended for staff certification and soon it is planned to conduct a mass certification, more than 50k people.
I have made a small application that emulates user behavior (browser) in order to pass a test and check with how many simultaneous users the server can handle.
Specificity is that users will have access to the application on a schedule and a surge of simultaneous requests to start the test is possible.
So, I set up my app that it emulates the test start situation 10 times at an every second and after 3 - 3.5 thousand starts, the connect function crashes with the error # 10061 (WSAECONNREFUSED).
It is not a timeout, but an active refusing of the connection.
Do I understand correctly that because of the load, the apache thread that calls accept function should simply do not have time to work out all the connections, and after the queue of such “underconnects” overflows, they start to interrupt with a failure?
There are 350 threads configured in apache, and up to 30 php-cgi processes in mod_fcgid.
These are the most optimal settings chosen by practical consideration.
Reducing and increasing them leads to the fact that connections begin to refuse at earlier stages.
I found the backlog setting in apache, but the I see that nothing changes as I increase this number.
I don’t understand what can I do. What can be done to make the server to survive this burst of simultaneous connections?

[James Blond]

Please post your mpm settings and the fcgid settings.#

Any error in the windows event log? Errors in the apache error log file?