AdaBoost Second Learner Error

AdaBoost is run on four training points, all initialized with weight $\frac{1}{4}$. The first learner $G_1$ misclassifies exactly one of the four points. Each weight is then multiplied or divided by $\sqrt{(1 - \text{err})/\text{err}}$ in the usual way.

The second learner $G_2$ misclassifies exactly one different point (one that $G_1$ had classified correctly). Compute the weighted error rate $\text{err}_2$ of ${}_{}$.