Sorry, I missed the the second part of your question. You don't measure the onsager cavity radius, since it's not an experimental observable. Rather, you set it when using the Onsager model. Remember that it's just a model (and a simple one) that has little to do with the true solute/solution system, though it can give good results for some systems.
There are a few ways to set it - see the first link in my previous message. In general, when doing QM calculations, it's either setup automatically or manually. In Gaussian you can request the program to calculate the volume of a system first (keyword: volume) in gas phase and then use the recommended cavity radius value in the next step when using Onsager model.
If you are using Gaussian, you can use the VOLUME keyword in the SCRF part of your route section in order to get the actual size of your cavity, which is the smallest sphere that can contain your solute defined as vdw overlapping spheres centered on each atom/group of atoms (some models include H atoms as a whole topological group)