residual diagnostic for multilevel model

Dear all,

I'm doing residual diagnostic for multilevel models with continuous outcome in Stata. I'm using the following commands. The level 2 residuals look sort of flat, and I'm wondering what the next steps may be if I conclude that level 2 residuals are not quite normally distributed. Any suggestion would be much appreciated! Thanks!

mixed y c.x1##i.time || id: time, cov(un) nolog reml

predict lev1, residuals
predict lev2a lev2b, reffects

histogram lev1, xtitle(Level 1 residuals) normal
histogram lev2a, xtitle(Level 2 residuals) normal
histogram lev2b, xtitle(Level 2 residuals) normal