The initial document count is very slow since the server seems to be scanning the whole collection.

To reproduce: try to load a big collection, the first load will take seconds. Moving to next page afterwards (which does not involve re-counting) is almost instantaneous.

I'd expect the first page load to be as fast as the next page since the same amount of documents are loaded.

Possible fixes:

• Make the count independent from the find and only report the result in the ui when is ready.
• Use the id field as hint if query is empty: countDocuments({}, {hint: '_id_'}

  ). For what i could test this speeds up the count a tons, the server seems to do an index scan in this case.