In the case with the disc, the mass is distributed fairly uniformed around the observer at the axis so he would barely experience any attraction at all.
That's just not so.
Assuming Newtonian style gravity:
Suppose the gravitational source is a disk in the x-y plane centered on the origin, with a uniform density &rho and radius a.
Suppose we have a point observer on the positive z axis measuring gravity. Let z represent his z-coordinate, and let his mass be m.
For an area element of the disk dA located a distance r from the origin, its mass is &rho r dr d&theta and its distance from the observer is sqrt(z^2 + r^2), so the z-component of the gravitational force on the observer by this area element is:
dF = -G m z / (z^2 + r^2)^(3/2) &rho r dr d&theta
Integrating over the disk:
&int dF = &int -G m z &rho r / (z^2 + r^2)^(3/2) dr d&theta
= (-G m z &rho) &int d&theta &int r / (z^2 + r^2)^(3/2) dr
= (-G m z &rho) (2&pi) (-1) (r^2 + z^2)^(-1/2) | r = 0..a
= (-2&pi G m z &rho) (1 / z - 1 / sqrt(z^2 + a^2))
= -2&pi G m &rho (1 - z / sqrt(z^2 + a^2))
Due to symmetry, the x and y components of the gravitational force are 0, so the magnitude of the gravitational force experienced by our axial observer is:
F = 2&pi G m &rho (1 - z / sqrt(z^2 + a^2))
Notice that:
F -> 2&pi G m &rho as z -> 0
so that it is
not true that F -> 0 as z -> 0.
Plugging in some numbers for comparison:
F(z = a) = 2&pi G m &rho (1 - 1 / sqrt(2)) = 2&pi G m &rho * 0.2929
F(z -> 0) = 2&pi G m &rho
So we see that as our observer approaches the disk's surface, the gravitational force he feels increases up to about 3.414 times as strong as the force he felt when his distance from the disk was equal to its radius.
Taylor expanding around z = 0 shows that near the surface, deviation from the uniform field F = 2&pi G m &rho is second order in z, so when you're sufficiently close to the surface, you appear to be in a uniform gravitational field.
This is all exactly the same as the case for a uniformly charged plate in electrostatics.
For a disk with thickness t, because of the aforementioned near-uniformity, an observer near the surface of the cylinder (or on the surface of the cylinder, as long as he's not sinking
into the cylinder) will experience a gravitational force:
F = 2&pi G m &rho t
For the exact solution for an observer who is a height h above the surface of the cylinder, integrate F(z) for z in h .. h + t yielding
F(h) = 2&pi G m &rho (t + sqrt(h^2 + a^2) - sqrt(h^2 + (t + a)^2))
Plugging in h = 0 yields:
F(h = 0) = 2&pi G m &rho (t + a - sqrt(t^2 + a^2))
So again we see that an observer near (or on) the surface experiences a nonvanishing gravitational force. The total mass of the cylinder is &pi a^2 t &rho, so we can rewrite the above equation using M as the total mass of the cylinder:
M = &pi a^2 t &rho
&pi &rho = M / (a^2 t)
F(h = 0) = 2 G m M / (a^2 t) * (t + a - sqrt(t^2 + a^2))
Plugging in some numbers for comparison, suppose the thickness of the disk is 1 unit and the radius is 1000 units. Then, the gravitational force is approximately:
F(h = 0) = G m M (2 * 10^-6)
Which is the same gravitational force felt by an observer 707 units from a point gravitational source with the same mass as the cylinder. Hardly "barely any attraction at all".
For completeness, another example would be a thickness of 1 unit and a radius of 10^6 units. In this case, the equivalent observer would be 7*10^6 units from the point mass. Clearly the ratio between thickness and radius has a dramatic effect on things, but as pointed out before the force will never vanish, and an observer standing on (not in) the disk will always experience a greater gravitational attraction than an observer who is some distance away from the disk along the axis.
When a is much larger than t, the equation reduces to:
F = 2 G m M / a^2
An observer standing on a disk of negligable thickness and radius a will experience a gravitational force twice as much as an observer standing on a sphere with the same mass and radius.
This all assumes uniform thickness; in reality we would expect the disk (or cylinder) to be more dense in the center, this increasing the force felt by our axial observer.
Hurkyl