All possible (x,y,z) are x=y=[1,2,…,2k−1] and z=[1,2,…,2k], where k is any positive integer and […] denotes the least common multiple.
Clearly z>x,y. Suppose x=[1,…,a], y=[1,…,b], and z=[1,…,c], and without loss of generality assume a≤b. Furthermore, without loss of generality let a be maximal (that is, for any a′>a we have x=[1,…,a′]), let b be maximal, and let c be minimal. This necessarily guarantees that c>b and y∣z−y=x, so x=y and z=2x. By this maximality and minimality, this means a=b=c−1, and also 2v2(c)∤y but 2v2(c)−1∣y. This can only hold when c is some power of 2 (otherwise 2v2(c)<c would divide y.) Hence the solutions must be of the form x=y=[1,2,…,2k−1] and z=[1,2,…,2k], where k is any positive integer. Substituting back to check shows this clearly holds. Q.E.D.