Those logs look fine to me. We actually have many users indexing several millions of documents effortlessly so I'm really curious to know what's going wrong here. Trying on Linux directly is a good idea.