Repository size got bigger after cleanup :(

Manuel Aude March 22, 2018

A few weeks ago I saw the size was at around 800mb (we store the content of several sites in a single repo for management purposes). So, I ran BFG to clean it up and did a push force, pretty much just like in the documentation I found here

 

https://confluence.atlassian.com/bitbucket/reduce-repository-size-321848262.html

 

Now, I proceeded to check... my local repo got much smaller but the size reported at bitbucket didn't change.

 

Since it was far from 1gb and I was busy, I decided I had to prioritize other issues.

Now, today I checked and the size was 1.8gb! Only around 10 websites had been added since, and the total size was only a few megabytes (no more than 50mb), yet the repo got 1gb bigger.

 

So, I decided to run filter-branch command this time.

 

Since we store our sites by date and currently no websites from 2017 or older are being deployed to production (and they have been deleted with git rm already), I ran:

 

git filter-branch --tree-filter 'rm -rf 2017' --prune-empty HEAD
git for-each-ref --format="%(refname)" refs/original/ | xargs -n 1 git update-ref -d
git gc --prune=now
git push --all --force

 

And then proceeded to check the repo size... yet this time is shows 2.2GB? What? Why? Now it got into read-only mode, even though I have been trying to keep it under control, following your own documented proceedures.

 

Not being able to push into this repository is a big problem for me. I did back up the last with --mirror, so I will apply some urgent settings, but I don't understand why this happened and I really want to avoid this from happening ever again.

 

2 answers

0 votes
Joey Rick January 21, 2019

The pack sizing increase issue (after running BFG) happened to me and ended up being due to git version 2.18 era version pack issues.  A coworker using 2.19 did not have the issue.  I found the bug fix description in the 2.19.0 release notes - * In a recent update in 2.18 era, "git pack-objects" started producing a larger than necessary packfiles by missing opportunities to use large deltas.  This has been corrected.

0 votes
Boris Berenberg - Atlas Authority
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
March 22, 2018
Manuel Aude March 22, 2018

Uhm I hadn't. However, on one hand, that talks only about BFG, and I used filter-branch on the second situation. On the other hand, even though it talks about probable causes, I fear I don't see solutions being proposed.

I already proceeded to use my backup repository and redirected deployment settings to use it. I mainly just want to know how can I avoid this issue in the future.

Boris Berenberg - Atlas Authority
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
March 23, 2018

There is a comment that says "I'm running into the same issue with bitbucket as well. In the comments on this page it looks like you have to email bitbucket support and request an additional git gc on your repo." 

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events