Gentoo Archives: gentoo-scm

From: "Robin H. Johnson" <robbat2@g.o>
To: gentoo-scm@l.g.o
Subject: [gentoo-scm] repo layout & graft / split-history
Date: Mon, 01 Nov 2010 04:40:11
Message-Id: 20101101043958.GA1264@orbis-terrarum.net
In Reply to: Re: [gentoo-scm] meeting followup: repo layout by Donnie Berkholz
1 On Fri, Oct 29, 2010 at 09:28:34PM -0500, Donnie Berkholz wrote:
2 > With 4K blocks, that works out at roughly 500 MB (CVS) to 2 GB (SVN,
3 > Git) of inode overhead. I have a hard time imagining people so hard up
4 > for disk space that they can fit the whole git repo but can't find
5 > another 1.5 GB.
6 I'd like to ask the embedded arch folk how they feel about that size
7 proposal. I know some of the MIPS team were running 2.1GB SCSI drives,
8 and complaining already.
9
10 > Based on my current git conversion with a pack size of 1.7 GB, I suppose
11 > that means the total repo in a git world could vary from ~2 GB all the
12 > way up to ~4 GB.
13 There is one oversimplification here, that the pack will actually be
14 that large...
15
16 I'm playing with graft points, and the tree as it stands, with no
17 history (single commit), is a single 66MiB pack. I've included some more
18 numbers for going further back in history:
19
20 Command I've used to generate the numbers below:
21
22 # ( cd gentoo-x86-git-authors.git/ ; git fast-export ${START}..HEAD ) | \
23 ( rm -rf trim ; mkdir trim ; cd trim ; git init ; \
24 time git fast-import --stats ; \
25 du .git ; \
26 time git gc --aggressive ; \
27 du .git )
28
29 End point: Tue Oct 26 15:43:01 2010 +0000
30 Pack sizes by starting point:
31 Tue Oct 26 15:43:01 2010 +0000: 71016 KiB (single commit)
32 Fri Oct 01 00:25:21 2010 +0000: 81760 KiB
33 Wed Sep 01 02:20:31 2010 +0000: 92636 KiB
34 Sun Aug 01 00:45:49 2010 +0000: 103248 KiB
35 Fri Jan 01 00:00:39 2010 +0000: 182440 KiB
36 ...
37
38 Approximately 11MiB/month of growth in the packfile.
39
40 Is the packfile of the kernel sources an acceptable size?
41 It's presently ~800MiB. If we start with zero or minimal history (6
42 months maybe). This gives us a fairly small tree...
43
44 Still needed with the above:
45 - Document _very_ well how to graft history back together if you want
46 the full history.
47 - Prevent full history from being pushed back.
48
49 --
50 Robin Hugh Johnson
51 Gentoo Linux: Developer, Trustee & Infrastructure Lead
52 E-Mail : robbat2@g.o
53 GnuPG FP : 11AC BA4F 4778 E3F6 E4ED F38E B27B 944E 3488 4E85

Replies

Subject Author
Re: [gentoo-scm] repo layout & graft / split-history Caleb Cushing <xenoterracide@×××××.com>
Re: [gentoo-scm] repo layout & graft / split-history Mike Auty <ikelos@g.o>
Re: [gentoo-scm] repo layout & graft / split-history Donnie Berkholz <dberkholz@g.o>