I want to prove that $f(x) = \sum_{n=1}^{\infty} \frac{\log(1 + \frac{x}{n})}{n}$ is not uniformly convergent on the whole of $\mathbb{R}$.
My idea was to make the argument that if $f(x)$ were uniformly convergent on $\mathbb{R}$, then $\exists N\in\mathbb{N} $ such that $|f(x) - f_n(x)|<\epsilon $ for any $\epsilon > 0$ and any $x \in \mathbb{R}$ when $N\leq n$. Let $f(x)-f_n(x) = \sum_{k = n+1}^{\infty}\frac{\log(1 + \frac{x}{k})}{k}$. Choose $\epsilon = 1$ and $x = k\cdot e^k$. Then, $\left|\sum_{k = n+1}^{\infty}\frac{\log(1 + \frac{x}{k})}{k}\right|= \left|\sum_{k = n+1}^{\infty}\frac{\log\left(1 + \frac{k\cdot e^k}{k}\right)}{k}\right| > 1$, which contradicts the proposition that we can make $\bigl|f(x) - f_n(x)\bigr|$ smaller than any epsilon by choosing $N$ large enough.
Somehow this argument feels faulty. Can I really choose $x$ in such an arbitrary way? $k$ is after all not a constant, but $x$ might need to be?
Any insights into this would be appreciated.
You have the right idea. Here is another way to use the same idea.
Let $f_n(x)=\sum_{j=1}^n\frac {\log (1+x/j)}{j}$ for $x>-1$ (...so $\log (1+x/j)$ is defined when $j=1$).
Let $f(x)=\lim_{n\to \infty}f_n(x).$
Let $\|f-f_n\|=\sup_{x>-1}|f(x)-f_n(x)|.$
Then directly from the definition, $f_n$ converges uniformly to $f$ on $(-1,\infty)$ iff $\lim_{n\to \infty}\|f-f_n\|=0.$
For $m>1$ let $x_m=m(e^m-1).$ We have $f_m(x_m) -f_{m-1}(x_m)=1.$
For any $n$ we have $$\|f_{n+1}-f\|+\|f-f_n\|\geq$$ $$\geq |f_{n+1}(x_{n+1})-f(x_{n+1})|+|f(x_{n+1})-f_n(x_{n+1})|\geq$$ $$\geq |f_{n+1}(x_{n+1})-f(x_{n+1})+f(x_{n+1})-f_n(x_{n+1})|=$$ $$=|f_{n+1}(x_{n+1})-f_n(x_{n+1})|=1.$$
Therefore $\max (\|f_{n+1}-f\|, \|f-f_n\|)\geq 1/2.$
So for every $n$ there exists $n'\geq n$ (...namely $n'=n$ or $n'=n+1$) such that $\|f-f_{n'}\|\geq 1/2.$ Therefore $\|f-f_n\|$ does not converge to $0$ as $n\to \infty.$