The probability distribution of the sum of two independent variables with probability distributions requires a convolution integral:
fz(z) = Integral from -inf to inf of (fx(z-y)*fy(y)dy) or the same integral with x and y switched.
There will be four delta functions that have to be multiplied together and integrated. The first term after distributing will be
integral of 1/3 δ(z-y) * 1/2 * δ(y)dy = 1/6 δ(z) Multiply coefficients and δ(y) only exists at y = 0, so plug y = 0 into other term in order to get z function.
Good luck!