The gravitational redshift is not the same as a Doppler shift, which is due to the relative motion between light source and observer. The gravitational redshift is due to the interaction between the electromagnetic field and the gravitational field. The latter is described, within standard GR, as the curvature of space-time. It is not possible to completely disentangle the two effects, because this "disentanglement" is a frame-dependent issue. The math is the only way to make this clear. So here it is:
In eikonal approximation you can calcuate the redshift as follows. You solve the eikonal equation
$$g^{\mu \nu} \partial_{\mu} \psi \partial_{\nu} \psi.$$
The characteristics of this partial differential equation give the direction of wave propagation, and thus
$$k_{\mu}=\partial_{\mu} \psi$$
is the frequency-wavenumber fourvector of the light wave. The eikonal equation implies that this is a light-like vector
$$k_{\mu} k^{\mu}=0,$$
and the "light rays" in the sense of geometrical optics given as the null-geodesics of space-time.
A "pointlike observer" is uniquely and covariantly specified by giving his four-velocity, ##u^{\mu}##, which is a time-like vector. In natural units, where ##c=1##, you have ##u_{\mu} u^{\nu}=1##. Then the frequency of the light measured by this observer at his location in space-time is
$$\omega_{\text{obs}}=k_{\mu} u^{\mu}.$$
That's it, and this formula includes both the gravitational redshift for light emitted from a heavy object (like the sun) and the Doppler effect due to the relative motion between the observer and the light source. If the observer is at rest relative to the light source, you have a pure gravitational redshift; otherwise gravitational redshift and Dopplereffect are intermingled.
Another special gravitational redshift is the Hubble redshift. Here you can define this specific redshift, because you can define it as the redshift of light emitted from a distant source as measured by a comoving observer who is defined as being at rest relative to the large-scale coarsegrained "cosmic substrate". Note that for such an observer the cosmic substrate and thus the light source are uniquely defined to be at rest. This is because of the symmetry of the underlying space-time model, the Friedmann-Lemaitre-Robertson-Walker pseudometric, which is the realization of the cosmic principle, i.e., the assumption that the local laws on the large-scale coarse-grained level are everywhere and always the same and that spacetime is thus maximally symmetric.