Hmm, this could get complicated...
My first thought was well let's just look up "open tube resonance". But that would be for the air column inside your tube. My second though was, I think we're trying to ring the cylinders themselves, so we don't much care about the air column. If that's the case you have to calculate a bunch of stuff along the lines of material stiffness, young's modulus, and such. Which is perhaps too much information.
Maybe the best way to go is experiment. (Tune and) Ring a cylinder, get the frequency, and then calculate what the others should be. If all the cylinders are the same material and size, I think your frequencies should scale linearly with length.
Depending on how much of a tin ear you have you may need to do some fine tuning. Removing material from the end is an easy way to raise the frequency. I think you can also remove material from the 1/3 nodal point in the middle to lower the frequency -- that works with marimba-like bars at least.