Yes, another twist...
Just setting the displacement for an ungapped guide will not allow you to consider friction coming from the guide. You are on the right track though. Use the Node/CNode restraint definition and then set your coefficient of friction on that restraint (and displacement on the CNode).

If you set this up properly you should be able to confirm support action when both the pipe & support are moving differing amounts. Once again, close review of results can confirm your model.

(I tend to discount friction on supports not carrying weight. Weight bearing restraints have a more consistent normal load so more consistent friction. With your gapped guide, there will be some free sliding before the pipe engages the pipe and friction actually begins on this lateral restraint.)