Thanks a lot Lew I think it actually helps

There are of course several factors to take into account as very well pointed out. What I sort of lock on to is that one should be careful about the fit, this since I understand as the cage can quite easily be compressed and then actually hinder the rollers from turning.
Let me explain a bit more about my intended use as it may give some more info.
It started some time ago when making a light duty exchangeable live center for my Unimat clone lathe. The idea and drawing came at the time from Bob Loaders book on Unimat accessories. It was made as a housing with a "hole" of the same diameter as the existing tail stock but with three bearing balls in the bottom. By making different and exchangeable tips this little live center became extremely useful, see pic below. So useful that I had in mind to do something similar for my C3 mini-lathe.

In the picture we see the housing at the back to the left. The tips are all steel apart from the two items in the front row to the left. Those are the original live center and the original dead center. A small wooden stand helps me keep them in order. They are all made on the small lathe.
For the slightly larger contraption I thought that using needle roller bearings would give a better turning motion and less heat build-up, and by trapping a thrust needle bearing at the bottom I would get a better bearing than just three balls.
So, I started off by getting a blank end arbor for the housing, two caged needle rollers with 15 mm ID and 22 mm OD, and a thrust bearing with 21 mm OD. The tips I intend to make from a good steel, close to silver steel but not hardened as my steel bits have proven to wear very little even without hardening.
So, here is my origin for the question. The tips will of course be made to a suitably close fit in the roller bearing and to run very close to the housing to prevent swarf as good as possible. The lubrication will be the same type as used in front hubs on cars, that is a high pressure, high temp, high quality bearing grease.
But back to the original quesstion, how tight a fit? As I understand it the fit ought to be a light push fit so the cage does not spin on its own, but not so tight that the rollers are stopped. Possibly one could use some loctite stuff to glue the bearings in place to secure them a bit more?
Am I completely up the wrong tree here? (And yes it is kind of an extension to the Stevensonīs small live center )
BR
/Peter