For the revolute joint, is there a Joint Load of type Moment applied to balance F*D? That would generate a zero pivot error in a Static Structural model unless you used Inertia Relief.
If the joint load was a Rotation of 0 degrees, then you could probe the reaction moment on the joint. That would solve without the zero pivot error.
In either case, the nodes around the cylindrical face get the same force vectors, which are mostly tangential in direction.
The cantilever flat face has force vectors on the nodes are mostly horizontal in direction.