In @PavelBoytchev’s solution, the flat part on which you move always stays down. In @prisoner849’s solution, you rotate in the tube and the flat part is in different positions.
I have not yet found a solution for the rotation in my problem Inaccurate adjustment when cylinder is deformed. Must see if I can learn something from these solutions. 