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

Re: Relocate issue

From: Ryan Schmidt <subversion-2006Q1_at_ryandesign.com>
Date: 2006-02-08 23:08:31 CET

On Feb 8, 2006, at 22:29, Wilson, Kalin (N-Scitor) wrote:

> 1. Created a new empty repository on the new machine svn create d:
> \svnrepos
> 2. dumped the old repository into the new repository svnadmin
> dump y: | svnadmin load d:\svnrepos (where y: is a network mapped
> drive to the other \svnrepos)
> 3. Told all users to update from old repository
> 4. shut down old repository server svnserve to ensure no more
> changes were commited
> 5. did one final synch of new to old repositories to ensure both
> were at same HEAD revision
> 6. told users to relocate working copies by executing TortoiseSVN
> relocate command on root of the working copy, changing the svn://
> URL to the new URL
> All this went fine with no errors from tortoise during the
> relocate. However, I tried to commit modifications from my working
> copy to the new repository and got a commit error with no
> explanation. I verified that problem from another users working copy.
>
> I relocated my working copy back to the old (original) repository
> and I was able to commit the changes.
>
> Did I do anything wrong? My guess is there is a synchronization
> issue with modified files within the working copy. But it may not
> be practical to require all users to commit all modifications
> before we have to move the repository.
>
> Id rather avoid having to checkout new working copies from the new
> repository.

You shouldn't have to check out new working copies or commit your
work before the relocation. It should just work, assuming the old
repository was successfully copied to the new one.

Loading a full dump into a new repository should set the new
repository's unique ID to be the same as the old one, which is
required for the relocation to work.

You told us how you initially dumped the data into the new
repository. You imply that during this import, users were still
working on the old repository, which would be why you did one final
synchronization after you finally shut off access to the old server.
What command did you use to synchronize? It seems to me that you
would have had to specify an incremental dump starting at the HEAD of
the new repository and ending at the HEAD of the old repository. Is
that what you did?

What exactly happened when you tried to commit? It failed with no
output at all? Are we talking about the Subversion command line
client or some GUI like TortoiseSVN? If a GUI, try the command line
client and see what it says.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Feb 8 23:09:55 2006

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