Matt Trost wrote:
> Ryan Schmidt wrote:
>> On Jan 10, 2008, at 09:56, Matt Trost wrote:
>>> Ryan Schmidt wrote:
>>>> On Jan 9, 2008, at 16:02, Matt Trost wrote:
>>>>> I recently experienced every developers' nightmare - the server
>>>>> crashed. I have my SVN repository back in place on a different
>>>>> server, and I luckily had both a full backup (hot copy) and the
>>>>> incremental revisions committed after the last hot copy. When I
>>>>> loaded these incremental revisions back into the repository it
>>>>> created transactions, but didn't commit them. I can't figure out
>>>>> how to get them committed so that I can get back to work.
>>>> Presumably you're using "svnadmin load" to load the incremental
>>>> revisions into the new repository? What's the exact command you're
>>>> using? And what's the error output, if any? Are the permissions on
>>>> the repository OK? What version of SVN do you have and on what OS?
>>> Yes - I'm using the "svnadmin load" command. The transactions were
>>> placed into the db directory in the repository
>>> (D:\SVNRepository\db\transactions). There are ten transactions
>>> waiting to be committed. I'm using SVN 1.4.5 on a Windows Terminal
>>> Server 2000 machine now - it was 2003.
>>> This is the script in the post-commit hook batch file that created
>>> the incremental revision file (as far as I can remember):
>>> REM do an incremental backup of the commit
>>> @ECHO OFF
>>> REM %1 = repository path, %2 = revision number
>>> SET REVISION=%2
>>> svnadmin dump D:\SVNRepository\ -r %REVISION%:%REVISION% >
>>> path-to-backup-directory\revision%REVISION% --incremental
>>> Maybe this hook script wasn't functioning properly and was creating
>>> bogus files.
>> You probably meant to pass the --incremental switch to svnadmin like
>> svnadmin dump D:\SVNRepository\ -r %REVISION%:%REVISION%
>> --incremental > path-to-backup-directory\revision%REVISION%
>> I think this just means you were getting full revision backups
>> instead of incremental ones, so this would have taken more time to
>> make and would have been bigger but should still have worked.
>>> I removed the last transaction and used the following command to try
>>> to reload it:
>>> svnadmin load path-to-repository < path-to-file\revision316
>>> This is the output/message that I got back:
>>> <<< Started new transaction, based on original revision 316
>>> * adding path : mpps ...svnadmin: File already exists:
>>> filesystem 'path-to-repository/db', transaction '306-10', path 'mpps'
>> So the file that is supposed to represent revision 316... actually
>> represents revision 306? Hmm.
> You might be right, Ryan. The incremental dump files are large, much
> larger than I was thinking they should be. Maybe I can start with an
> empty repository and load the last dump file (revision316). I'll give
> that a try.
I was able to reload the repository from the latest dump file, and it
put the revision at 1. I guess this is fine. I'm going to verify that
everything is there that should be. Thanks for your help.
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-01-10 22:33:55 CET