Need GC on remote

[justin.charbonneau]

Dear Bitbucket support team,

We have a repo that has over 2 GB (thus the warning). 

I wasn't quite 100% sure on the sequence of git commands to run to first delete select large files and clean the history. I was able to clean it following the instructions below and removed large CSV files from my repo delete_history_test. The repo is still 66MB and should be much smaller. Can you run the GC on that repo, so I have proof that these instructions work? After which I'll talk to my team to clean up our repo and I suppose we will need to trigger a GC next week. 

Dummy Exercise:

What I wanted to have is 

• main branch with checked-in large files.
• infected branch with still work to do (but everything committed and pushed to remote)

The large files exist in both the main and the infected branches, so I did:

• git checkout main
• git rm large_file.csv
• git commit -m "delete large file"
• git push
• git checkout infected branch
• git rm large_file.csv
• git commit -m "delete large file"
• git push

Then, I clean the history doing:

• git checkout main

• git filter-branch -f --prune-empty --index-filter "git rm --cached -f --ignore-unmatch large_file.csv" --

• git push -f
• git checkout infected branch
• git filter-branch -f --prune-empty --index-filter "git rm --cached -f --ignore-unmatch large_file.csv" --
• git push -f

I suspect after this point, all of the histories should be clean and I need to trigger the GC from the remote which I can't trigger. 

Can you trigger it to this repo delete_history_test so I can make sure it works?

Thanks,

Justin

[Theodora Boudale]

Hi Justin,

I ran a git gc for this repo and its size is 58.6 MB now.

Can you take a mirror clone of the repo and execute git count-objects -Hv in the mirror clone directory, and let me know what you see in the output?

Kind regards,
Theodora

[justin.charbonneau]

Hi Theodora, here's my output from that command:

How come the repo is still 58MB ? There is barely anything in the repo?

[Theodora Boudale]

Hi @justin.charbonneau,

In your first post, you mentioned that the repo has more than one branch.

1. Does your clone (where you executed git count-objects -Hv) have all branches locally?

If you haven't pulled all remote branches locally, then the local repo will be of a smaller size.

2. Could you please take a mirror clone of the repo delete_history_test

git clone --mirror <repo_url>

and then run git count-objects -Hv in that mirror clone, and let me know what the output is?

Kind regards,
Theodora

[justin.charbonneau]

Hi @Theodora Boudale

Initially yes it had more than one branch. In the meantime I had discovered that maybe I could use LFS to move those files there, and I tried to play around with it. At one point in time, the repo went down to 60 KB when I had first created another branch and installed LFS and added the gitattributes for CSV files, but after merging it, the repo went back to 66MB. I tried to reproduce but I couldn't. In the end I deleted the LFS file and tried to clean it up to bring it back to that initial state. 

Here's a screenshot of the commands you told me to run:

[Theodora Boudale]

Hi @justin.charbonneau,

Thank you for the info.

Since the mirror clone also shows a size of 8.96 KB, we will need to inspect the remote repo to see what takes up extra space.

I cannot access your repo in order to check that, we need a support ticket in order to be able to access a remote repo.

I see that the test repo delete_history_test belongs to your personal workspace, but there is no other repo there over 2 GB.
I also see that you are a member of a workspace on a paid billing plan.
Does the repo that has a size of over 2 GB belong to that workspace?

If so, what you can do is:

• talk with your admins and transfer delete_history_test repo to the workspace on the paid billing plan
• you can then create a ticket with the support team to look into the issue with the size. You can create a ticket via https://support.atlassian.com/contact/#/, in "What can we help you with?" select "Technical issues and bugs", then Bitbucket Cloud as product, and enter the workspace URL of the paid billing plan.

However, if delete_history_test repo is not related to a repo on that paid workspace and it's your personal repository, please feel free to let me know and I can create the support ticket for you.

Kind regards,
Theodora