Is language $L_2=\{w^*\,:\,w\in L_1\}$, where $L_1$ is regular, regular?
Intuitively it seems that it's not: the automaton accepting $L_2$ would most probably just be a loop made from an automaton accepting $L_1$ (it would accept a regular expression $L_1^*$). However, how would that automaton "remember" what was the last word read, and thus if all the repeating words all the same?
In other words it seems impossible to build an automaton that could tell $w_1\cdot w_1 \cdot w_1$ and $w_1 \cdot w_2 \cdot w_1$, for $w_1, w_2 \in L_1$ apart.
On the other hand, it seems impossible to proove that $L_2$ is not regular using the puming lemma - for $u \in L_2$, $u = w^n$, you could just pump $w$, and $u_2 = w^{n+k}\in L_2$