Use Robust Jarque Bera (RJB). The Robust Jarque–Bera (RJB) is the robust version of the Jarque–Bera (JB) test of normality. The Robust Jarque Bera (RJB) is in R Package ‘lawstat’. See the journal article: Uyanto, S. S. An Extensive Comparisons of 50 Univariate Goodness-of-fit Tests for Normality. Austrian Journal of Statistics, 51(3), 45–97. Retrieved from https://ajs.or.at/index.php/ajs/article/view/1279 (an open access, peer-reviewed journal free for everyone to read and download)
All hypotheis-tests are highly sensible to sample size. The higher the sample size, the higher the chance to reject the hypothesis, conclude no to normality and making type-I error.
I recommend you to instead visually check residual behavior from quantile-quantile plots (qq-plots) and residual vs predicted plots.