NaNs in regression summary statsmodel. Error with clustering standard errors?

Question

I\'ve encountered something I\'ve never seen before when using statsmodel\'s ols.

I have experimental panel data of water consumption on a monthly basis. I\'m running