2 ms·

I think the dupe-detector only checks against URLs that have already been lazy-loaded into memory since the last time the server was restarted.