Depending on the server, I'd say the client is either delayed or gets an HTTP error code in the 500 range as a response. If you're lucky, you're able to deliver a custom error page which is able to retrieve both IP and locale, but this would seem like handling the request in spite of being unable to handle it ... So, I'm not optimistic about that.
You could try to create a simulation for this kind of server load and see what happens. I doubt the server's documentation is detailed enough.
As I said, it depends on the server software. If the server provides a way to handle denied requests, you might be able to know. If you expect a server load of more than 1000 parallel requests, you should watch out for a pro solution using load balancing via clustered listeners and not go for the all-free, ad-sponsored host ...