Search for related literature where similar compounds are handled computationally. Alternatively, you can do benchmark study by yourself by choosing a similar molecule whose properties are known, and see which basis set and DFT functional (in case you're using DFT) gives best match. Also, Take a look into related books and papers (e.g: Chapter A Chemist's Guide To Density Functional Theory
) where detail discussion about DFT functional along with there performance are discussed.
While Na is not transition metal and have not much electrons it has no sence to use pseudopotential basis set for it. Without gaining anything one might just get additional (normally quite small) error from combination of basis sets.
I also see no point to start with CoupledCluster method at the very beginning of work. DFT with some double-zeta basis set will give you starting understanding.
Double-zeta quality basis set is senceless in combination with CCSD, and LANL2TZ that might have some sence for CCSD does not contain Na or anything else except of transition metals.
Nevertheless, advice to read articles about Lanl2DZ basisset is good. One can find that 1) initial developer (1977) was Dunning who after that switched to his own sets. 2) Papers of Hay and Wadt are mainly focused on heavy elements and transition metals.
Surely there exist people who calculate even pure organic with Lanl2DZ, however its rather specific choice.