How can I fork a huge repository when even a single branch is larger than the 2GB size limit?

[k414]

Attempts at forking from the official Odoo repository (https://github.com/odoo/odoo.git), even only the branch of interest (10.0) ends up creating a repository of 2.3 GB.  This ends up exceeding Bitbucket's size limit effectively making the fork useless.

What steps can be taken to retain the original historical branch, allow for syncing future upstream patches, and allow commits back to the forked repo with customized modifications (the whole point of the fork)?

Thanks in advance!

[Julius Davies [bit-booster.com]]

Can you get the original upstream "odoo.git" repo on github to shrink themselves?  That would probably be the best option.

Barring that, depends if you're willing to let all of the git commit-ids change in your fork or not.  Of course if all the commit-ids get changed, it becomes much harder to keep things in sync with upstream (which would still have the original commit-ids).   Normal "repo" shrinking tools like "bfg-repo-cleaner" alter every commit-id because they change the contents of historical commits (to remove unnecessary large files). This causes a domino-effect on the commit-ids, since every commit-id after that historical commit will change.  This is a fundamental property of the Merkel-Tree data structure that Git is built on.

If you need the commit-ids to stay the same as upstream (github.org), then the only option is to use something other than bitbucket.org to host your fork.

On-prem Bitbucket does not have size limits.