If documents are very large but don't contain large strings, the BSON output in the logs is not truncated. This leads to huge logfiles, as well as potential slowdown which would result in more logging, more slowdown, etc.