Of course there will be a phase difference.
For a hypothetical, let's start with two absolutely identical GPSDOs with antennas at the same place, near as possible, and identical length cables. To the extent that we manage to get everything exactly identical, the 10 MHz oscillators ought to be exactly in phase.
If we lengthen the cable of one of the oscillators, its phase will be delayed by the amount of time it takes the signal to travel the extra coax length.
10MHz is a wavelength of about 30 meters. If we add a half wavelength of coax to the cable between one of the antennas and its oscillator, that will delay the signal by a half-wavelength, and put the oscillator 180 degrees out of phase from its twin. To be more precise, we'd have to divide by the velocity factor, so 18.75 meters of coax would be about a half of a 10MHz wavelength after accounting for a velocity factor of 0.8.
You say you've got about 45 feet of difference in the coax lengths. That's approximately 15 meters. That's on the same order of magnitude as a 180 degree phase shift.
There may be other sources of phase shift besides the cable length, as edpalmer42 discusses. But the cable length alone should cause an easily observable phase difference.