There may be phase variations $ge lambda/4$ across the diameter of each mirror in any laser cavity?

In high quality optical systems, $lambda/4$ is often considered the maximum allowed peak to peak variation of the optical path difference allowed in the beam that forms an image. Many also use a maxiumum allowed rms (root mean square) variation of 0.07 waves.

Since ideally a laser cavity would be diffraction limited, so that it produces a diffraction limited beam, then it seems reasonable to put a $lambda/4$ limit on allowed phase variations (or optical path difference) on optics inside the cavity. In practice, laser manufacturers would typically specify much tighter specifications on mirrors: $lambda/20$ would be common. They are trying to get as high a beam quality and power as possible out of their laser. Since I'm not a laser expert, I can't comment on how much phase variations affect the extraction of power from within the laser cavity.