You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Gustavo Niemeyer <ni...@conectiva.com> on 2004/01/22 14:20:44 UTC
dump & load
Hello folks,
I have a small question about "svnadmin load" behavior. Any help will
be appreciated.
Suppose I have a repository with 1000 revisions. Then, I "svnadmin
dump" that repository into two parts of 500 revisions. Even though the
second part is non-incremental, is it possible to load it over the
first part and get the same repository with 1000 revisions back?
Thanks!
--
Gustavo Niemeyer
http://niemeyer.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by plasma <pl...@pchome.com.tw>.
On Thu, Jan 22, 2004 at 01:25:43PM -0600, Ben Collins-Sussman wrote:
> Oh, you want *replication* from one server to another.
>
> You should come into IRC sometime. People have reinvented solutions to
> this over and over... we have 'svnpush' in our source tree (which could
> be run as a post-commit hook.) svk does this too. And someone recently
> wrote a bourne script to do this as well, runnable from a cron job.
svm (which is in SVN::Mirror that can be found on CPAN) will do if all
you want is replication. 'svk mirror' and 'svk sync' simply call
SVN::Mirror to do the actual works.
plasma
==========================================================
�����k�ʡ��n�Χܵk�����K�O�e���p
http://edm-prg.epaper.com.tw/click.php?ad_code=57495
==========================================================
PChome���~~���֤ѤѦ� \*^o^*//
http://love.pchome.com.tw/
==========================================================
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by Gustavo Niemeyer <ni...@conectiva.com>.
> > I'm thinking about something slightly different: maintaining two
> > repositories side-by-side, one of them being the working repository,
> > and the other being for read-only access. From time to time I'd take,
> > for example, 10.000 revisions from the bottom of the working copy,
> > and move them to the top of the read-only version.
>
> Oh, you want *replication* from one server to another.
>
> You should come into IRC sometime. People have reinvented solutions to
Erm.. I'm there most of the time. :-)
> this over and over... we have 'svnpush' in our source tree (which could
> be run as a post-commit hook.) svk does this too. And someone recently
> wrote a bourne script to do this as well, runnable from a cron job.
This is very interesting, but not really what I have in mind in this
case. I'm interested in cutting the bottom of the working repository to
maintain it below some given size (10GB, 20GB, 50GB, I don't know). As I
said, this is just an idea right now. I'll have to think further about
the advantages and consequences.
--
Gustavo Niemeyer
http://niemeyer.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by Ben Collins-Sussman <su...@collab.net>.
On Thu, 2004-01-22 at 13:11, Gustavo Niemeyer wrote:
> I'm thinking about something slightly different: maintaining two
> repositories side-by-side, one of them being the working repository,
> and the other being for read-only access. From time to time I'd take,
> for example, 10.000 revisions from the bottom of the working copy,
> and move them to the top of the read-only version.
Oh, you want *replication* from one server to another.
You should come into IRC sometime. People have reinvented solutions to
this over and over... we have 'svnpush' in our source tree (which could
be run as a post-commit hook.) svk does this too. And someone recently
wrote a bourne script to do this as well, runnable from a cron job.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by Gustavo Niemeyer <ni...@conectiva.com>.
> > Humm.. btw, what about if the first dump had revisions 1 -> 500, and
> > the second one had 500 -> 1000. Would it be possible to ignore
> > revision 500 of the second dump and append them together again?
>
> Sure, you could do that by removing r500 from the 2nd dumpfile and then
> appending the 2nd dump to the 1st dumpfile.
>
> But this is why we have 'svnadmin dump --incremental', no?
>
> $ svnadmin dump -r0:500 > dumpfile1
> $ svnadmin dump --incremental -r501:1000 > dumpfile2
>
> Then later on
>
> $ cat dumpfile2 dumpfile2 > dumpfile
> $ svnadmin load repos < dumpfile
Yes, I understand this process already.
I'm thinking about something slightly different: maintaining two
repositories side-by-side, one of them being the working repository,
and the other being for read-only access. From time to time I'd take,
for example, 10.000 revisions from the bottom of the working copy,
and move them to the top of the read-only version.
This is just an idea right now, but I may actually do that in the
future to avoid handling a 26GB+/46000+ repository everytime I have
to do a backup, or dump/load, or whatever.
--
Gustavo Niemeyer
http://niemeyer.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by Ben Collins-Sussman <su...@collab.net>.
On Thu, 2004-01-22 at 12:54, Gustavo Niemeyer wrote:
> Humm.. btw, what about if the first dump had revisions 1 -> 500, and
> the second one had 500 -> 1000. Would it be possible to ignore
> revision 500 of the second dump and append them together again?
Sure, you could do that by removing r500 from the 2nd dumpfile and then
appending the 2nd dump to the 1st dumpfile.
But this is why we have 'svnadmin dump --incremental', no?
$ svnadmin dump -r0:500 > dumpfile1
$ svnadmin dump --incremental -r501:1000 > dumpfile2
Then later on
$ cat dumpfile2 dumpfile2 > dumpfile
$ svnadmin load repos < dumpfile
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by Gustavo Niemeyer <ni...@conectiva.com>.
> Not likely. The first dumped revision of your second dumpfile will
> look like an "add" of your entire tree -- that's not gonna work out
> well, considering that at load-time, you'll be trying to perform adds
> of paths that already exist in the repository. You'd basically need
> to tweak that first revision (revision 501?) of your second dumpfile
> so that it only contains the paths that were actually changed in that
> revision.
Humm.. btw, what about if the first dump had revisions 1 -> 500, and
the second one had 500 -> 1000. Would it be possible to ignore
revision 500 of the second dump and append them together again?
--
Gustavo Niemeyer
http://niemeyer.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by Gustavo Niemeyer <ni...@conectiva.com>.
> Not likely. The first dumped revision of your second dumpfile will
> look like an "add" of your entire tree -- that's not gonna work out
> well, considering that at load-time, you'll be trying to perform adds
> of paths that already exist in the repository. You'd basically need
> to tweak that first revision (revision 501?) of your second dumpfile
> so that it only contains the paths that were actually changed in that
> revision.
This was exactly my doubt. Thanks!
--
Gustavo Niemeyer
http://niemeyer.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: dump & load
Posted by "C. Michael Pilato" <cm...@collab.net>.
Gustavo Niemeyer <ni...@conectiva.com> writes:
> Hello folks,
>
> I have a small question about "svnadmin load" behavior. Any help will
> be appreciated.
>
> Suppose I have a repository with 1000 revisions. Then, I "svnadmin
> dump" that repository into two parts of 500 revisions. Even though the
> second part is non-incremental, is it possible to load it over the
> first part and get the same repository with 1000 revisions back?
Not likely. The first dumped revision of your second dumpfile will
look like an "add" of your entire tree -- that's not gonna work out
well, considering that at load-time, you'll be trying to perform adds
of paths that already exist in the repository. You'd basically need
to tweak that first revision (revision 501?) of your second dumpfile
so that it only contains the paths that were actually changed in that
revision.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org