| 1 |
After talking with some git developers at the Summer of Code mentor |
| 2 |
summit, I resumed work on getting gentoo-x86 into git. Today I got the |
| 3 |
first conversion to git from cvs working, using cvs2svn's cvs2git |
| 4 |
backend. I attached the config files I used. Here was the process: |
| 5 |
|
| 6 |
- Start with an rsync of the gentoo-x86 cvs repo |
| 7 |
(see http://anoncvs.gentoo.org/). |
| 8 |
- `mkdir CVSROOT` in the same level as gentoo-x86 |
| 9 |
- cvs2svn --options=cvs2svn-git.options | tee cvs2svn.log |
| 10 |
- mkdir gentoo-x86-cvs2git |
| 11 |
- cd gentoo-x86-cvs2git/ |
| 12 |
- git init |
| 13 |
- cat ../cvs2svn-tmp/git-blob.dat ../cvs2svn-tmp/git-dump.dat \ |
| 14 |
| git fast-import | tee git-fast-import.log |
| 15 |
- time git repack -a -d -f -l --window=50 | tee git-repack.log |
| 16 |
(took ~25 minutes) |
| 17 |
- rsync -avzP -e ssh .git/ dev.gentoo.org:public_html/gentoo-x86.git/ |
| 18 |
|
| 19 |
Here's some stats on the conversion: |
| 20 |
|
| 21 |
Total CVS Files: 341121 |
| 22 |
Total CVS Revisions: 2035092 |
| 23 |
Total CVS Branches: 0 |
| 24 |
Total CVS Tags: 0 |
| 25 |
Total Unique Tags: 0 |
| 26 |
Total Unique Branches: 0 |
| 27 |
CVS Repos Size in KB: 1412266 |
| 28 |
Total SVN Commits: 591101 |
| 29 |
First Revision Date: Thu Jul 27 17:35:42 2000 |
| 30 |
Last Revision Date: Sun Oct 26 15:43:43 2008 |
| 31 |
------------------ |
| 32 |
Timings (seconds): |
| 33 |
------------------ |
| 34 |
22743 pass1 CollectRevsPass |
| 35 |
38 pass2 CleanMetadataPass |
| 36 |
0 pass3 CollateSymbolsPass |
| 37 |
184 pass4 FilterSymbolsPass |
| 38 |
3 pass5 SortRevisionSummaryPass |
| 39 |
0 pass6 SortSymbolSummaryPass |
| 40 |
222 pass7 InitializeChangesetsPass |
| 41 |
2535 pass8 BreakRevisionChangesetCyclesPass |
| 42 |
13238 pass9 RevisionTopologicalSortPass |
| 43 |
60 pass10 BreakSymbolChangesetCyclesPass |
| 44 |
218 pass11 BreakAllChangesetCyclesPass |
| 45 |
187 pass12 TopologicalSortPass |
| 46 |
457 pass13 CreateRevsPass |
| 47 |
0 pass14 SortSymbolsPass |
| 48 |
0 pass15 IndexSymbolsPass |
| 49 |
176 pass16 OutputPass |
| 50 |
40062 total |
| 51 |
|
| 52 |
|
| 53 |
It took about 12 hours in total, running on a brand-new consumer-level |
| 54 |
box (no RAID, etc). It might run as much as 3x faster if I put the whole |
| 55 |
repo into a ramdisk beforehand, because that ought to remove a lot of |
| 56 |
the pass1 time (reading ,v files). The other place that took a lot of |
| 57 |
time, pass9, seemed mostly CPU-bound. |
| 58 |
|
| 59 |
If you want to grab a copy of the tree and check it out, you can. If |
| 60 |
you're a dev, don't clone over ssh because it hits the server really |
| 61 |
hard. (This remains to be solved.) Here's the place: |
| 62 |
|
| 63 |
git clone http://dev.gentoo.org/~dberkholz/gentoo-x86.git |
| 64 |
|
| 65 |
The repo is around 900M, including all history, thanks to git's pack |
| 66 |
compression. That's roughly equivalent to the size of a cvs checkout |
| 67 |
with no history. |
| 68 |
|
| 69 |
One obvious thing to fix for a final version is adding the author map |
| 70 |
into the config files so we get real names for people. That won't be |
| 71 |
very hard -- we should be able to just pull everything from ldap. |
| 72 |
|
| 73 |
A clear next step is to compare a native cvs checkout with a cvs |
| 74 |
checkout of the git repo through `git cvsserver`. Does anyone want to |
| 75 |
help out and do this? |
| 76 |
|
| 77 |
-- |
| 78 |
Thanks, |
| 79 |
Donnie |
| 80 |
|
| 81 |
Donnie Berkholz |
| 82 |
Developer, Gentoo Linux |
| 83 |
Blog: http://dberkholz.wordpress.com |
Archives