The long baseline neutrino experiments, T2K and NOvA, have taken significant amount of data in each of the four channels: (a) $ u_mu$ disappearance, (b) $bar u_mu$ disappearance (c) $ u_e$ appearance and (d) $bar u_e$ appearance. There is a mild tension between the disappearance and the appearance data sets of T2K. A more serious tension exists between the $ u_e$ appearance data of T2K and the $ u_e / bar u_e$ appearance data of NOvA. This tension is significant enough that T2K rules out the best-fit point of NOvA at $95%$ confidence level whereas NOvA rules out T2K best-fit point at $90%$ confidence level. We explain the reason why these tensions arise. We also do a combined fit of T2K and NOvA data and comment on the results of this fit.