Find and resolve HTTP request bottleneck

We have a system which is checking proxy server uptime and experiencing an unexpected result after a recent server change. We need a server administrator that is comfortable with tuning PHP, Apache, PHP-FPM, MySQLi, and CentOS core. You as well need to be fluent with PHP multi-curl to understand how everything is working, in order to successfully resolve this issue for us. Please note this issue is server config related, not software related. The system uses PHP multi curl to select 20,000 nodes from the proxies and then process them in batches of 10,000. Requests are sent to our internal judge on the same server which returns information about the node that made the request. The problem we are experiencing is that we are limited to the 10,000 batch. As we increase the batch amount, it's actually successfully checking less proxies. For example, with the 10,000 batch, the system might return 3,000 proxies that are currently up. If we increase that to a 20,000 batch, the system might find 500 proxies that are currently up, off the same exact data set which is surely not correct On our previous server we could run batches of 100k without experiencing this type of issue but unfortunately, that administrator is no longer with us. Any idea of what could be causing this issue for us? We are ready for immediate assistance!