If we want to calculate $E(\mathbb{Z}/p\mathbb{Z},\mathbb{Z})$, i.e. equivalence classes of short exact sequences $\mathbb{Z}\rightarrow E\rightarrow\mathbb{Z}/p\mathbb{Z}$, we have $\mathbb{Z}/p\mathbb{Z}\cong\mathrm{Ext}(\mathbb{Z}/p\mathbb{Z},\mathbb{Z})\cong E(\mathbb{Z}/p\mathbb{Z},\mathbb{Z})$.
To get the explicit representative, we form the pushout as shown in the diagram: $$ \begin{array}{llllllllllll} \mathbb{Z}&\overset{\times p}{\rightarrow}&\mathbb{Z}&\rightarrow&\mathbb{Z}/p\mathbb{Z}\\ \downarrow\times a &&\downarrow &&{=}\\ \mathbb{Z}&\rightarrow&E&\rightarrow&\mathbb{Z}/p\mathbb{Z} \end{array} $$
The pushout is calculated as $\mathbb{Z}\oplus\mathbb{Z}/((a,p))$. When $(a,p)=1$, we have $ax+py=1$, so the invertible matrix $\begin{bmatrix}x & y\\-p & a\end{bmatrix}$ will take $(a,p)$ to $(1,0)$, which shows $E\cong\mathbb{Z}$. And $(1,0)$ is taken to $(x,-p)$, which is then equivalent to $-p$ in $E\cong\mathbb{Z}$, so the lower column is again a $\mathbb{Z}\overset{\times p}{\rightarrow}\mathbb{Z}\rightarrow\mathbb{Z}/p\mathbb{Z}$ whenever $(a,p)=1$. But for different $a=1,2,...,p-1$, the sequence should be different, as $\times a$ are different elements in $\mathrm{Ext}(\mathbb{Z}/p\mathbb{Z},\mathbb{Z})$, which is a contradiction.
Surely I am wrong somewhere, but I just cannot find it out. Can anyone help me? Thanks.