> scan rules :
> the new files were downloaded, but some options were not 
> taken into account (deletion of -*.pdf), the index.html

Did you interrupt "suddenly" (without waiting for pending 
downloads) the mirror? This may have left some files 
behind, not referenced anymore.. are the pdf links 
rewritten to "external" (http://) urls after that?
if these files are not references anymore due to "sudden" 
stop, you'll have to delete them (might be feasable through 
find files / older than..)