[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: very large revision file - very small amount of data

From: Julian Foad <julian.foad_at_wandisco.com>
Date: Fri, 12 Feb 2010 16:42:21 +0000

Kevin mailed me today to say:
> Appreciate the response. This is the same repository that Trent is
> working on, and he has explained the issue to me. This group created a
> directory which now has 50k plus directories in it causing the
> problem. We have asked them to restructure the repository so there are
> not so many directories in one place.

The existence of a directory containing 50000 entries, which take around
40 or so bytes each to store, explained the 2 or 3 MB revision file size
they were seeing.

- Julian

On Thu, 2010-02-11 at 14:11 +0000, Julian Foad wrote:
> Konstantin Kolinko wrote:
> > 2010/2/11 Kevin Longfellow <klongfel_at_yahoo.com>:
> > > Just wondering if someone can shed some light on why a revision file
> > is 18 times greater than the amount of checked out data for that
> > revision? I'm using Subversion 1.6.5 but not an upgraded and packed
> > repository. I'm currently testing that to see if it will make any
> > difference and so far it looks like the savings is minimal.
> > >
> > > The issue here could be misuse of source control, and I am trying to
> > find out more about their use. This group has an automated test
> > application that will create 1400+ revisions of around 3MB revision
> > files per day. With the inability to delete this can't go on for to
> > long.
> [...]
> > > There is 164 KB of data in the checkout but the revision is 3.0MB?
> >
> > 1. This is a question for the users@ list
> > 2. You are committing Test_99820 but checking out Test_174313
> > 3. There was Test_99820.zip
>
> A very similar query was posted on a forum recently:
> <http://subversion.wandisco.com/forums/2-general-support-and-troubleshooting/178-repository-disk-space-analysis-tools.html>.
>
> In that case, that user, Trent, had a similar situation and was
> investigating quite deeply (One thing he said was, '[...] I do a "cat"
> of the current and previous revisions and then run "xdelta" on those
> files under the assumption that will approximate how much space that
> revision occupies.' And, 'I picked a revision, and got the involved
> files... they totaled 336k and the delta from the last version was about
> 90k. But the revision file in the repository was over 2 gigs!')
>
> I suggested that he might like to contact the dev list to get further
> help in debugging the problem, but he has not done so yet.
>
> Kevin might find it useful to compare notes or talk to Trent.
>
> - Julian
>
>
Received on 2010-02-12 17:43:02 CET

This is an archived mail posted to the Subversion Dev mailing list.