You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by Edi Stojicevic <es...@debianworld.org> on 2008/03/18 16:43:13 UTC

Tool to test two repositories after a dump/load

Hi,

I have 103 repositories that I dumped/loaded to a new server using this
command :

svnadmin dump repo | ssh -C svnadmin load repo

Everything went fine but when I checked the size of everything I have a
difference of near 50Gb missing on the target.

On the source I have a Debian server with :

    subversion version  : 1.3.2-5 
    Filesystem          : xfs

On the target Redhat with :

    subversion version  : 1.4.2-2.el5 
    Filesystem          : ext3


For now I checked two repositories only and I have the same number of
revisions and the same number of files which is nice for now.
But I would like to know if there is a tool or something to provided
some more tests on the data to be sure that nothing has been corrupted
like md5 checksums by example.
Do you know any tool, perl module to do that ? 

Any other advice is welcome ...

Cheers,

-- 
. ''`.  (\___/) E d i   S T O J I C E V I C
: :'  : (='.'=) http://www.debianworld.org 
`. `~'  (")_(") GPG: 0x1237B032
  `-

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: Tool to test two repositories after a dump/load

Posted by Edi Stojicevic <es...@debianworld.org>.
* kmradke@rockwellcollins.com <km...@rockwellcollins.com> [2008-03-18 15:46:35 -0500] wrote :

[...]

> > I don't know of a tool, but the size difference could be normal. The 
> > repository format was changed in Subversion 1.4 to be more efficient. 
> > Your 50GB difference may be the result of this greater efficiency.
> 
> You will also be missing any "incomplete" transactions and lock files.
> (Although neither of these would normally be 50G in size)
> 
> I would suspect the 1.3->1.4 repo format changes also, but it would be
> interesting to know total repo sizes.
> 
> I've gained a couple of GB reloading old 20GB repos, but it really depends
> upon the files that were archived.

There is a lot of mp3 and flash (I didnt do the site ;) files and I saw a lot of files copied three times in .svn directories...

> Kevin R.

-- 
. ''`.  (\___/) E d i   S T O J I C E V I C
: :'  : (='.'=) http://www.debianworld.org 
`. `~'  (")_(") GPG: 0x1237B032
  `-

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: Tool to test two repositories after a dump/load

Posted by km...@rockwellcollins.com.
Ryan Schmidt <su...@ryandesign.com> wrote on 03/18/2008 
03:32:21 PM:
> On Mar 18, 2008, at 11:43, Edi Stojicevic wrote:
> 
> > I have 103 repositories that I dumped/loaded to a new server using 
> > this
> > command :
> >
> > svnadmin dump repo | ssh -C svnadmin load repo
> >
> > Everything went fine but when I checked the size of everything I 
> > have a
> > difference of near 50Gb missing on the target.
> >
> > On the source I have a Debian server with :
> >
> >     subversion version  : 1.3.2-5
> >     Filesystem          : xfs
> >
> > On the target Redhat with :
> >
> >     subversion version  : 1.4.2-2.el5
> >     Filesystem          : ext3
> >
> >
> > For now I checked two repositories only and I have the same number of
> > revisions and the same number of files which is nice for now.
> > But I would like to know if there is a tool or something to provided
> > some more tests on the data to be sure that nothing has been corrupted
> > like md5 checksums by example.
> > Do you know any tool, perl module to do that ?
> >
> > Any other advice is welcome ...
> 
> I don't know of a tool, but the size difference could be normal. The 
> repository format was changed in Subversion 1.4 to be more efficient. 
> Your 50GB difference may be the result of this greater efficiency.

You will also be missing any "incomplete" transactions and lock files.
(Although neither of these would normally be 50G in size)

I would suspect the 1.3->1.4 repo format changes also, but it would be
interesting to know total repo sizes.

I've gained a couple of GB reloading old 20GB repos, but it really depends
upon the files that were archived.

Kevin R.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: Tool to test two repositories after a dump/load

Posted by Ryan Schmidt <su...@ryandesign.com>.
On Mar 18, 2008, at 11:43, Edi Stojicevic wrote:

> I have 103 repositories that I dumped/loaded to a new server using  
> this
> command :
>
> svnadmin dump repo | ssh -C svnadmin load repo
>
> Everything went fine but when I checked the size of everything I  
> have a
> difference of near 50Gb missing on the target.
>
> On the source I have a Debian server with :
>
>     subversion version  : 1.3.2-5
>     Filesystem          : xfs
>
> On the target Redhat with :
>
>     subversion version  : 1.4.2-2.el5
>     Filesystem          : ext3
>
>
> For now I checked two repositories only and I have the same number of
> revisions and the same number of files which is nice for now.
> But I would like to know if there is a tool or something to provided
> some more tests on the data to be sure that nothing has been corrupted
> like md5 checksums by example.
> Do you know any tool, perl module to do that ?
>
> Any other advice is welcome ...

I don't know of a tool, but the size difference could be normal. The  
repository format was changed in Subversion 1.4 to be more efficient.  
Your 50GB difference may be the result of this greater efficiency.



---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org