Re: cvsup

["Simon 'corecode' Schubert" <corecode@xxxxxxxxxxxx>]

Vincent Stemen wrote:
> On 2008-01-15, Simon 'corecode' Schubert <corecode@fs.ei.tum.de> wrote:
> > Yes.  There are a couple of mirrors which serve the cvs repo via rsync. 
> > For instance use chlamydia.fs.ei.tum.de.  There are more, I guess.
>
> Thanks.  
>
> I just checked the list, but most of the sites say they only mirror
> _Daily snapshots and official ISOs_.  The rsync link to the one you
> mentioned, *chlamydia.fs.ei.tum.de*, does take you to a web page that
> says they have the cvsup collections, but they do not provide the full
> path to access them via rsync.

I guess you're expected to find out with rsync yourself :)

You know that you can get a listing with rsync?

sweatshorts % rsync chlamydia.fs.ei.tum.de:: 
               dflysnap        DragonFlyBSD daily snapshots
dragonfly-cvs   DragonFly CVS repository
freebsd-cvs     FreeBSD CVS repository
netbsd-cvs      NetBSD CVS repository
openbsd-cvs     OpenBSD CVS repository
sweatshorts % rsync chlamydia.fs.ei.tum.de::dragonfly-cvs 
               drwxr-xr-x         512 2007/05/07 23:38:22 .
-rwxrwxr-x         321 2003/07/15 06:17:23 fixme
drwxrwxr-x        1536 2008/01/16 12:33:43 CVSROOT
drwxrwxr-x         512 2006/06/28 22:24:29 dfports
drwxrwxr-x         512 2007/04/06 22:03:43 doc
drwxrwxr-x         512 2007/12/21 06:39:21 site
drwxrwxr-x         512 2004/06/18 13:23:48 src.sys.alpha.old
drwxrwxr-x        1024 2008/01/14 13:39:27 src

> > Rsync is very slow when you tag a repo, because all the files change, but 
> > only one line.  It's not really optimized for this task.  But this doesn't 
> > happen very often, though.
>
> That is interesting.  The rsync docs claim to be very fast and say this
>
>     The rsync remote-update protocol allows rsync to transfer just the dif-
>     ferences between two sets of files across the network connection, using an
>     efficient  checksum-search  algorithm ...

Well, of course in general it is fast.  But CVS repos are a special case, 
consisting of hundreds of thousands of small RCS text files, which only 
get one line added when the repo is tagged.  A general solution is bound 
to perform worse than a special case solution.

> I ran a test transfering 8 text files over the internet that are each about
> 450K and the performance seems pretty good when I changed just one line in each
> file.  In my test I told rsync to use ssh, so the transfer was also encrypted.

CVS repos are different:  Imagine 100000 files with an average size of 
4-8KB.  I think rsync has to do 100000 network roundtrips to negotiate on 
the changesets to be sent.

> Total bytes sent: 31162
> Total bytes received: 26842

Considering the fact that you only changed one line per file, that's quite 
a lot.  Plus it doesn't say how many round trips it was needing.

> As you can see, it took 9 seconds for a full download, but only less than 1.5
> seconds to update them.  That seems reasonably fast to me.  Is cvsup really
> faster than that?  I am sceptical that it could be much faster.

Yes, cvsup is way faster in such a case.  Maybe not necessarily for 8 
files, but for 800 for sure.  It pipelines communication, so that there is 
no need to wait for immediate replies, thus saving network roundtrip times.

cheers
  simon

--
Serve - BSD     +++  RENT this banner advert  +++    ASCII Ribbon   /"\
Work - Mac      +++  space for low €€€ NOW!1  +++      Campaign     \ /
Party Enjoy Relax   |   http://dragonflybsd.org      Against  HTML   \
Dude 2c 2 the max   !   http://golden-apple.biz       Mail + News   / \