Git Garbage collection doesnt seem to fully work
Asked Answered
R

1

21

I'm a little confused as how to completely clean out my garbage...

git count-objects -v -H

warning: garbage found: ./objects/pack/gc_7174754666377259454.idx_tmp
warning: garbage found: ./objects/pack/gc_7174754666377259454.pack_tmp
warning: garbage found: ./objects/pack/pack-f5b13f50fe2e4d773028c51f547822e6f2fe720b.bitmap
count: 0
size: 0 bytes
in-pack: 32986
packs: 1
size-pack: 44.14 MiB
prune-packable: 0
garbage: 3
size-garbage: 41.20 MiB

So that implies to me I have 41 megs of garbage in my repo?

git gc --prune=now --aggressive

Counting objects: 32986, done.
Delta compression using up to 8 threads.
Compressing objects: 100% (31610/31610), done.
Writing objects: 100% (32986/32986), done.
Total 32986 (delta 23902), reused 9080 (delta 0)

And when I run count objects again i still have the same output with

    size-garbage: 41.20 MiB

Do I just manually delete the garbage files? At least one is quite plump at the very least.

12/02/2014  02:06 PM                 0 gc_7174754666377259454.idx_tmp
12/02/2014  02:06 PM        43,195,455 gc_7174754666377259454.pack_tmp
               2 File(s)     43,195,455 bytes
               0 Dir(s)  502,905,999,360 bytes free
Roborant answered 2/12, 2014 at 19:30 Comment(1)
see if you have better luck with git repack -adUnderclothes
M
27
C:\Users\VonC\prog\git\git>git log -Ssize-garbage|more

This show the size-garbage output has been introduced in commit 1a20dd4 by Nguyễn Thái Ngọc Duy (pclouds) for git 1.8.3 (May 2013)

size-garbage: disk space consumed by garbage files, in KiB

count-objects: report how much disk space taken by garbage files

Also issue warnings on loose garbages instead of errors as a result of using report_garbage() function in count_objects()

This garbage cleaning tip section mentions:

To bring the repo size down the bare minimum, you need both the following commands (neither command by itself does the whole job).
Also Note the lowercase "a" on the "repack", which says you want to blindly discard unreachable objects instead of keeping them as loose objects.

git repack -adf     # kills in-pack garbage
git prune           # kills loose garbage

So try again the git count-objects -v -H after applying both commands.


Looking at the git repack man page, jthill adds in the comments:

I prefer the big-A option:

"Same as -a, unless -d is used.
Then any unreachable objects in a previous pack become loose, unpacked objects, instead of being left in the old pack."

Linus Torvalds argues that -f like gc's --aggressive is much overused -- so much so he suggested yanking the documentation for it.
(in 2007)
(-f is for --no-reuse-delta)

That means a more efficient combination might be:

git repack -Ad      # kills in-pack garbage
git prune           # kills loose garbage
Maggy answered 17/2, 2015 at 14:26 Comment(3)
I prefer the big-A option, "Same as -a, unless -d is used. Then any unreachable objects in a previous pack become loose, unpacked objects, instead of being left in the old pack." Linus Torvalds argues that -f like gc's --aggressive is much overused -- so much so he suggested yanking the documentation for it.Hereat
@Hereat Interesting. I didn't know about that 2007 thread. I have included your comment in the answer for more visibility.Maggy
Thank you for that link, @jthill. That was an incredible interesting and educational read.Fear

© 2022 - 2024 — McMap. All rights reserved.