Results 1 to 2 of 2

Thread: Migrate SVN repository from dapper to lucid

  1. #1
    Join Date
    Sep 2007
    Beans
    68
    Distro
    Ubuntu 10.04 Lucid Lynx

    Question Migrate SVN repository from dapper to lucid

    Hey everyone,

    I've been pulling my hair out on this for several hours now so I really hope you can help. I have a (very old) svn repositoy on an ubunta dapper machine. I want to switch to another machine with lucid running. I simply want to "copy" the whole repository to the new machine. I've tried several ways, but the most succesful one was to do a dump on the old machine, copy the file to the new machine and do a load there. The repository has about 28.000 revisions. At revision 3130 it stopped with an error saying some path was not in UTF-8. I tried running a UTF-8 migrationtool on the dapper machine but this told me all files were UTF-8 compatible already. I am really stuck, even a two and a half hour google session didn't provide any answer.

    edit: more specific, the error I get is:
    svnadmin: Path 'beginofpath?\218tabase' is not in UTF-8
    The end should have been /database instead of ?\218tabase. When I look in the revision on the dapper machine this path shows a square at that location, so this is wrong also.
    When I try to open the dumpdile in gedit I get an errormessage saying:
    Could not open the file /home/usrname/repo.svndump using the Unicode (UTF-8 ) character encoding
    When I change the encoding I get the same error for that encoding.
    Last edited by HandleWithCare; September 28th, 2010 at 03:11 PM. Reason: more specific error details

  2. #2
    Join Date
    Sep 2007
    Beans
    68
    Distro
    Ubuntu 10.04 Lucid Lynx

    Arrow Re: Migrate SVN repository from dapper to lucid

    In reply to myself (since no one else does ), here's an update.
    I have managed to move around this problem by doing partial dumps an loads. By doing this I was able to single out the revisions that caused the trouble. Luckily, the two revision that were causing all the hassle contained no binary data so I was able to edit those revisions in a text editor. There were some unrecognized characters, must have gone wrong during commit at that time (probably about 5 years ago). After that fix the revisions would load and after two hours of dumping and loading (ok, this starts to sound weird), all was good and we have a new svn server wich is extremely fast compared to the old one.
    I will not mark this thread as solved since the actual problem hasn't been fixed, I merely worked around it. Good ideas on how to prevent this in the future are still welcome.

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •