Why is remote repository size increasing after force push?

[Iris Seidinger _Reflekt_]

Hi there,

I am currently on a mission of cleaning up our repo, which has grown over the past 7 years to over 10 gb. I am migrating files to git lfs, use bfg repo cleaner and git-filter-repo to clean up. I set up a new repo where i only pushed one branch of the old repository and step by step do force pushes that clean up more things from history.

What I expect: a force push of a smaller repo size should also decrease the size on the remote, so that the bitbucket UI's "Retrieve size" should show approximately the same size.

What actually happens: after every force push, the respository size increases, sometimes by 500 mb sometimes by 3gb. The remote already shows a size of 7gb again even though a fresh checkout and my local copy both show sizes under 3gb.

In short: my local repository keeps decreasing in size but the remote keeps increasing.

What is going wrong? Do I have to run a git gc on the remote? is that even possible & recommended?

Any help and insights appreciated :)

[Tyler T]

If you are working with Bitbucket Server, the instance should take care of garbage collection automatically. You can trigger garbage collection manually using the directions found here: https://confluence.atlassian.com/bitbucketserverkb/how-to-perform-a-manual-garbage-collection-on-a-repository-867194296.html

If you continue to have issues, please contact support.

[Iris Seidinger _Reflekt_]

Hi @Tyler T , thanks a lot for the response! We triggered "git repack -Adln" but omitted the keep-unreachable flag though as its not available on the ancient git version that our bitbucket server is currently on (2.9) (Our repository is at this point only used for testing so it felt safe to omit that flag)

This reduced the repo size from 7.6GB to 1.5GB! thats a lot! Now i am wondering why the server did not trigger any garbage collection automatically? How can we ensure that the issue is not reappearing? We are on Version "Atlassian Bitbucket v5.0.10"

Would be great to get some further insights

[Tyler T]

It might be most helpful to have our support team look at your specific instance to identify why it is not happening automatically. You can open a support ticket here: https://support.atlassian.com/contact/#/