git: dangling blobs

冷暖自知 提交于 2019-12-17 18:36:40

问题


I recently ran git fsck --lost-found on my repository.

I expected to see a couple dangling commits, where I had reset HEAD.

However, I was surprised to see likely over several thousand dangling blob messages.

I don't believe anything is wrong with my repository, but I'm curious as to what causes these dangling blobs? There's only two people working on the repository, and we haven't done anything out of the ordinary.

I wouldn't think they were created by an older version of a file being replaced by a new one, since git would need to hold onto both blobs so it can display history.

Come to think of it, at one point we did add a VERY large directory (thousands of files) to the project by mistake and then remove it. Might this be the source of all the dangling blobs?

Just looking for insight into this mystery.


回答1:


Last time I looked at this I stumbled across this thread, specifically this part:

You can also end up with dangling objects in packs. When that pack is repacked, those objects will be loosened, and then eventually expired under the rule mentioned above. However, I believe gc will not always repack old packs; it will make new packs until you have a lot of packs, and then combine them all (at least that is what "gc --auto" will do; I don't recall whether just "git gc" follows the same rule).

So it's normal behavior, and does get collected eventually, I believe.

edit: Per Daniel, you can immediately collect it by running

git gc --prune="0 days"



回答2:


I was really impatient and used:

git gc --prune="0 days"



回答3:


Whenever you add a file to the index, the content of that file are added to Git's object database as a blob. When you then reset/rm --cached that file, the blobs will still exist (they will be garbage collected the next time you run gc)

However, when those files are part of a commit and you decide later to reset history, then the old commits are still reachable from Git's reflog and will only be garbage collected after a period of time (usually a month, iirc). Those objects should not show up as dangling though, since they are still referenced from the reflog.



来源:https://stackoverflow.com/questions/9955713/git-dangling-blobs

标签
易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!