Given two short exact sequences of modules $$0\to N\to P\to R\to 0$$ $$0\to R\to Q\to M\to 0$$ Denote their extension classes by $e_1\in \mathrm{Ext}^1(R,N)$, $e_2\in\mathrm{Ext}^1(M,R)$.
Suppose $e_1\neq 0,e_2\neq 0$, is it possible that $e_2\cup e_1=0\in \mathrm{Ext}^2(M,N)$?
(On the one hand, it is plausible to have zero-divisor for cup product, on the other hand, not sure why two complicated extension may cancel to a trivial extension by splicing.)