Two vectors are orthogonal if their scalar product is zero. Also, since we have a subspace of dimension 1 (the line) in a space of dimension 3 (R^3), the orthogonal complement has dimension 2.
This means that we're looking for a basis of the 2-dimensional subspace of R^3 defined by the property
[tex]l^{\bot} = \left\{(x,y,z) \in \mathbb{R}^3\ :\ (2, -6, 9)\cdot (x,y,z) = 0\right\}[/tex]
By definition, this means we want
[tex]2x-6y+9z=0[/tex]
and we can deduce
[tex]2x-6y+9z=0 \iff z = \dfrac{-2x+6y}{9}[/tex]
We obviously have 2 degree of freedom here (we're building a base of a 2-dimensional space): we can write a generic vector of [tex]l^{\bot}[/tex] as
[tex]\left(x, y, \dfrac{-2x+6y}{9}\right) = \left(x, 0, \dfrac{-2x}{9}\right) + \left(0, y, \dfrac{6y}{9}\right) = x\left(1,0,\dfrac{2}{9}\right) + y\left(0, 1, \dfrac{2}{3}\right)[/tex]
In other words, we just proved that any vector of [tex]l^{\bot}[/tex] can be written as a certain linear combination of vectors (1,0,-2/9) and (0,1,2/3), which means that they are a base of [tex]l^{\bot}[/tex]