I think most issues with hive.blog are because the new code doesn't interoperate with the existing nodes correctly, so they have to take the front-end servers and official node offline to avoid network split. Last few days before HF are pretty much ironing out the last remaining bugs via live testing...