Large repository size even after removing all files and git history

[Matej Murín]

Hi guys,

We exceeded the allowed 4GB size of the repository, so I opted to revert some commits and remove the large files.

That said, I removed all the large files and I left only the codebase in the repository which now has 10s of MB in total. However, the size of the Bitbucket repo remained at 2.43 GB. I thought it might be because of some files still stored in the history, so I removed the whole history as well, but that didn't affect the size either.

I'm reading now that garbage collector might need to be run on the repo. Is there a way I can run it myself or does the Bitbucket team need to run it?

Thank you

[Patrik S · Accepted solution]

Hello @Matej Murín ,

and welcome to the Community!

The garbage collection (GC) is indeed necessary to remove any dangling objects left behind after you clean up the repository. The GC is automatically executed from time to time in the repository based on internal heuristics, but we from support can help on manually running a GC for you.

I went ahead and checked the repositories you have access to and found one that matched the size you put in the description. The GC on that repo was completed successfully, reducing its size from 2.4GB to 2.2MB!

Would you please confirm if everything looks good on your end now?

If you were actually referring to a different repository, please share the masked name of it with us so we can trigger a GC for that repository as well.

Thank you, @Matej Murín !

Patrik S

[Matej Murín]

Wow, amazing!!! 😃 Working like a charm, thank you so much @Patrik S !

Just for the future reference - how often are the automatic GC clean ups ran? Perhaps once a day or it depends heavily on the case and the heuristic evaluation?

Thanks again

[Patrik S]

Wonderful! Thank you for the update and happy to hear it worked :) 

The automatic GC heavily depends on internal heuristics evaluation, such as loose objects and the size of the packfile, so there's not a specific "scheduled" time for it to run. It will be triggered when those heuristics reach a threshold.