For the person who is about to report this as a duplicate, "no, this isn't a duplicate".
So I am trying to prove the following two (related) problems:
Suppose $K$ contains a primitive $p$th root of unity. Let $L=K[x]/(x^p-\pi)$ where $\pi$ is the prime in $\mathcal{O}_K$. Then $L$ is a cyclic totally ramified extension of $K$.
Show that if $s\in G(L/K)=:G$ such that $\langle s\rangle=G$, then $s\in G_i$ and $s\notin G_{i+1}$ where $i=e_L/(p-1)$ where $e_L$ is the absolute ramification index of $L$. (In other words, show $i=e_L/(p-1)$ is the first jump index)
Let $e_K$ be the absolute ramification index of $K$, and let $n$ be a positive integer prime to $p$ and $n< pe_K/(p-1)$. Let $y$ be an element of valuation $-n$. Define $$ S(x)=x^p-x-y $$ and let $L=K[x]/(S(x))$. Show that $G_n=G(L/K)$ and $G_{n+1}=\{1\}$.
Essentially, my end goal here is understanding the following paper: http://www.jstor.org/stable/2372667 (for those interested)
$x^p-\pi$ is Eisenstein and irreducible, thus $L/K$ is a degree $p$ extension, i.e. cyclic. It's totally ramified since $x^p-\pi$ is Eisenstein. By exer 2.3.c in Serre, $G_{i+1}=\{1\}$. Consider the chain of extensions $\mathbb{Q}_p\subseteq \mathbb{Q}_p(\zeta)\subseteq K\subseteq L$. By definition, $$e=v_L(p)=[L:K]e_{K/\mathbb{Q}_p[\zeta]}[\mathbb{Q}_p(\zeta):\mathbb{Q}_p]v_{\mathbb{Q}_p}(p)=p(p-1)e_{K/\mathbb{Q}_p[\zeta]},$$ so $$ v_L(\zeta\pi^{1/p}-\pi^{1/p})=1+v_L(\zeta-1)=1+pe_{K/\mathbb{Q}_p[\zeta]}=1 + i $$ so $G_{i}$ is non-trivial, i.e. $G_i=G(L/K)$.