Uploading to a bucket has never had a practical size limit. Uploading to one of your server drives did: a file over 100 MB was refused part-way through, with an error that did not say why. A video export or a database dump is well over that.

What changed

A large file is now sent in pieces. Each piece is checked in on the server as it arrives, the file is assembled there, and it only appears under its final name once the last piece is in — so nobody on the server ever sees a half-written file.

A dropped connection is no longer a restart

If the connection fails mid-way, the next attempt asks the server how much it already holds and continues from there instead of starting again. On a flaky office Wi-Fi that is the difference between a 4 GB upload finishing and never finishing.

This applies to the file browser and to BucketPilot Drop. Nothing to configure.