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

Re: Memory size in 1.1.x

From: Ben Collins-Sussman <sussman_at_collab.net>
Date: 2004-11-28 05:13:02 CET

On Nov 27, 2004, at 5:22 PM, Sigfred Håversen wrote:

> There have been some earlier posts regarding amount of memory used
> during
> import of large repositories. As I understand it there have been
> commits (in
> apr as well as in Subversion) dealing with this issue. But a "svn up"
> or "svn
> co" still uses a good deal of memory (105 MB and 74MB respectively).
> Is this
> to be expected?

Yes, it is expected and well-known that the client's memory usage is
potentially unbounded for commits; the more items sent in a commit,
the more memory the client will use, although the memory growth is by a
tiny constant factor.

The design of 'svn commit' and 'svn import' has the client holding a
list of committed objects in memory.

>
> The imported repository has 5900 directories, 51932 files,

52000 files multiplied by a tiny constant? Yeah, that would still be a
measurable amount of memory during 'svn import'.

But are you really getting 100 megs of client memory when doing a
checkout of 52000 files? That surprises me somewhat.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Nov 28 05:14:14 2004

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