If by "best approach" you mean reasonably general (in terms of variables right until the last step), and reasonably rigorous, then here's my approach. I'd divide the operation into two stages:
simulate this circuit – Schematic created using CircuitLab
Stage 1 is a simple resistor potential divider, with the following relationship between \$V_G\$ and \$V_H\$:
$$ V_H = V_G\frac{R_2}{R_1+R_2} $$
Stage 2 is a classic non-inverting amplifier. Op-amp OA1's output may be considered a voltage source, meaning that the load R5 at its output can be disregarded. I'm of course assuming that it won't draw more current than the op-amp can source, and that any current it does draw will not diminish the output swing of ±10V.
The relationship between stage 2 input \$V_H\$ and output \$V_{OUT}\$ is:
$$ V_{OUT} = V_H\left(1 + \frac{R_3}{R_4}\right) $$
We combine those two equations into one, to obtain a general equation relating \$V_{OUT}\$ and \$V_G\$. Here I'll do that by substituting \$V_H\$ from the first equation into the second:
$$ V_{OUT} = V_G\frac{R_2}{R_1+R_2}\left(1 + \frac{R_3}{R_4}\right) $$
There's only one unknown here: \$R_3\$. All other resistances are known, and input and output conditions are:
$$
\begin{aligned}
V_G = +5V \\ \\
V_{OUT} = +10V \\ \\
\end{aligned}
$$
I'm setting output \$V_{OUT}\$ to its maximum possible value, with the understanding that this represents saturation, and a further increase in \$V_G\$ will not result in \$V_{OUT}\$ increasing beyond that.
Now it's just a question of plugging in all known values and solving for R3.