New Repository Exceeding Limit - Repository is ONLY 107MB

Question

A newly created Bitbucket repository 'thepls-dev' shows a 2.6GB repo size when we brought the size down to just over 100 MB.  You can see the initial commit here:

Writing objects: 100% (23260/23260), 106.95 MiB | 713.00 KiB/s, done.

We specifically removed large files from the tree/commit before even adding the files to our commit.  I cloned the master branch and looked at the file size and it's 107MB in size.  Why is the repo size set to standards of two decades ago?  These are private repositories where many sites if not all client sites exceed 2 GB.  Shouldn't include resources?  That's why we excluded them through .gitignore.  I would gladly pay $20-100 USD/month to boost the maximum repository size.

Daniil Penkin · Answer

That seems bizarre to me that you have to trigger garbage collection manually, it seems like it should be automated or triggered whenever the 2GB limit has been reached to ensure it is not a false-positive (on top of the normally scheduled garbage collection – assuming that already exists).

It is automated, however it doesn't trigger on every single push – because it is quite expensive operation. Moreover, when GC kicks in it may run with different modes: more or less aggressive. Especially when the repository becomes huge, GC takes significant time, and given that we don't know how much garbage is in repo before we collect it, we should be careful to not to run GC again and again if it doesn't make any value. I believe this was the case with your repository.

Cheers,
Daniil

Nitro Interactive · Answer

Hi Daniil,

Despite my previous tone, I appreciate it.  Attempting to file a support request lead me to the support forum.  Unfortunately, I do not know your system as you do, so I apologize for filing my request in the wrong place.  The history contains all of one push, but perhaps the initial commit was the cause.  That seems bizarre to me that you have to trigger garbage collection manually, it seems like it should be automated or triggered whenever the 2GB limit has been reached to ensure it is not a false-positive (on top of the normally scheduled garbage collection – assuming that already exists).  Thanks again, frustration can be all-consuming.

Kind regards,

Steve Giorgi

Daniil Penkin · Answer

Hello @Nitro Interactive,

Repository size includes entire history of your repository, not just its latest state (working directory). So if you have a large file somewhere in the history, it will still contribute to the overall repository size.

Why is the repo size set to standards of two decades ago?

The main reason here is performance of Git on large repositories. In fact, it is generally recommended to keep your repository under 1Gb. If you have to check in large files, you might want to make use of Git LFS (large file storage) – it was designed exactly for this reason.

I would gladly pay $20-100 USD/month to boost the maximum repository size.

Again, the reason for a limit is performance of Git which is a known by-design issue, and there's no easy way to improve it. We can't make a trade-off here.

Are you guys going to fix this? We have been waiting on it all day. Absolutely pathetic. Why don't you offer some premium services so that we can get some assistance?

Community is not the best place to seek for help like this as it's not being actioned immediately. As mentioned in the Reduce repository size guide, a request should be filed  with our Support team. For now I found the repository in question and triggered garbage collection – repository size is now down to 110Mb.

Cheers,
Daniil