In this question "Relative homology of a retract: Help to understand a proof." for example. It is stated, among other things, that the border operator that takes a relative cycle from $H_n(X,A)$ to its border in $H_{n-1}(A)$ is actually the $0$ operator. By saying the following sequence is exact: $$0 \to H_n(A) \to H_n(X) \to H_n(X,A) \to 0.$$
But this means that a chain in $X$ that has a border contained in $A$ actually has no border. But why isn't a $\mathbb{R}^2$ and the the open disk a counterexample? Surely the open disk is a retract of $\mathbb{R}^2$, but it is easy to imagine a 1-chain in $\mathbb{R}^2$ that has a border in the disk and its border is not $0$.
(Or the closed disk and a smaller closed disk.)