Greetings,
A year late, I know right? I think fix is to slightly modify the original log file, then have Urchin re-run. Since I recalled that Urchin is smart enough to detect if a log has been looked at via looking at the leading bytes of the file, I simply created text file with a few repeated lines of:
127.0.0.1
www.mysite.com - [07/Nov/2012:00:00:00 -0500] "GET /" 301 - "-" "-" "-" - 1461
Called the file with these log lines "foo.log". I then uncompressed the original log file. In your case, you would end up with "file" since compressed file was "file.tgz." Then create a new log file:
cat foo.log uncompressed_logfile > new_logfile
Compress the log file you have created. Rename the new log file to that of the *original* log file, so that Urchin will reprocess as normal.
It feels like, by adding the nonsense log lines at the top, it is enough for Urchin to attempt to reprocess it. This has worked well for us, though I still don't know what the problem is at root.