How to reliably download/receive large files from the client browser?

Question

Browsers

The task is to accept large files from clients on Amazon S3 - they download archives of 1-2Gb and more.
We tried from a browser, using Evaporate JS to upload directly to Amazon S3 - random failures occur: for example, they set the download for the night, and by morning they find that it is stuck at 91%. Although waste downloads in parts and tries N times to send the pieces that did not pass. The Amazon side hangs, and the problem is specific to some IP addresses, it seems.
As a temporary crutch, we are thinking of receiving files on our server, and already sending them to S3.
There is an option to use nginx and its upload module , but this does not solve the problem of a broken connection and attempts to reload only parts of the file, and not the entire file.
Tell me proven reliable solutions with resuming in parts from the browser?

Answer 1

No way.
2 GB via http cannot be reliably pushed.