How does BitBucket handle Pull Requests for updates to LFS Blobs

[Phillip Demro]

Thanks for looking.  I am wondering how BitBucket handles the scenario below:

1. Repository "A" has a Git-LFS blob, "blob a"
2. Alice Forks Repository A
3. Alice makes an update to "blob a" and pushes
4. Alice opens a PR to Repository A
5. PR is merged into Repository A

In this scenario, would the LFS blob be automatically updated in repository A?

[Phillip Demro · Accepted solution]

Copy pasted from SO:

This is a known issue, see current limitations for Git LFS with Bitbucket. I also hope Bitbucket (and Github!) is going to improve this.

Consequences:

• LFS files must be synced between the fork and the upstream repo manually
• Forks will also use significant Large File Storage space, not just the upstream (there is a free storage quota - but after that is exhausted you will have to pay for each fork, ouch!)
• To copy LFS files back from the fork to the upstream, you will either need write access to the upstream repo or the upstream owner has to do it himself

How to sync the LFS files:

1. Cloning the fork repo for the very first time - it does not have any LFS files copied yet, so you must clone it without the LFS files (otherwise cloning would fail)

export GIT_LFS_SKIP_SMUDGE=y
git clone <address of your fork>

1. Synchronizing upstream -> fork (copying LFS files to your fork) - initially and each time you git fetch the upstream

git lfs fetch --all <upstream>
git lfs push --all <fork>

1. Synchronizing fork -> upstream (only the current branch) - each time after your PR got merged

git lfs fetch <fork>
git lfs push <upstream>

If there is a better workflow than this, please let me know. This is a serious limitation of the Bitbucket LFS handling.

https://stackoverflow.com/questions/55359067/what-is-the-workflow-for-git-lfs-with-forks