The very initial state of the relativistic heavy-ion collision is represented by coherent chromo-fields, which is commonly referred to as "Glasma" and some numerical simulations including quark production also work successfully. However, there is no calculation yet of the Glasma simulation with magnetic field, but this is strongly demanded in the realistic study of the Chiral Magnetic Effect (CME): the anomalous current and the associated charge separation are not a static phenomenon as sometimes investigated in Euclidean lattice QCD but they are very dynamical. In this talk I will look over technical and conceptual difficulties of such CME-Glasma descriptions and discuss some theoretical attempts.