> The usual wisdom is that if your old server and new server use the
> same architecture, same major OS version, same version of APR and
> Subversion, you can just copy the repository, but for anything more
> complicated, a dump and load is recommended. With a BDB repository,
> copying the repository to a different architecture shouldn't work
> at all, I believe. So the fact that yours is slightly working means
> you have an FSFS repository. I'm not sure if the issues you're
> seeing are due to such issues, but if possible, try to find a
> machine with the same specs as the one that died, put the
> repository on it, and run svnadmin dump on it. Then try to load
> that dumpfile into a new repository created on your new server.
Thank you for your response, sir! In the interim I have done an
experiment that leads me to believe that regenerating the repository
from a dump file would not help.
I moved the repository elsewhere for safekeeping:
# mv /usr/local/repositories ~/
I then created a brand new repository under the auspices of a
Subversion system on the new architecture:
# mkdir /usr/local/repositories/
# svnadmin create --fs-type fsfs /usr/local/repositories/TST/
# cd /usr/local/
# chown -R svn repositories/TST/
and tested the new TST repository ... I got the same error. I then
remembered to change the default passwd file and tried again, still
got the same error. :-)
If the general wisdom is that it's still worth trying to create a
dump and load it, I will do that. But it's going to be a few hours
work, so if you think the above experiment demonstrates that there's
no point, I'd just as soon not bother.
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].
Received on 2009-06-03 03:26:09 CEST