Hi,
I have to agree also that a diode across the solenoid is necessary. That's because when the transistor turns off the voltage across the solenoid reverses, so the diode in the darlington package can not shunt the current because it is connected the wrong way (and it needs to be that way for other reasons).
And yes, a diode of the same rating as the max current is usually used. That's because at the instant of turn off the current through the diode is the same as through the solenoid. Thus with 2 amps through the solenoid at turn off that puts 2 amps through the diode at turn off at least until the energy in the coil dissipates. How fast the energy dissipates depends on the inductance of the coil and the drop across the diode. With a large inductance it takes longer, and with a larger diode voltage drop it actually takes less time. Using two diodes in series means the energy dissipates in about half the time of one diode. Usually though you dont need two diodes unless you need to dissipate this energy quickly for faster solenoid de-energizing.
Most diodes have an average current rating and a peak rating. You really need the peak rating here not the average rating, but it would be good to check the data sheet. Some data sheets show data for one half cycle or something like that where it doesnt repeat for something like 100 cycles. You can probably use that as a guide. Most likely a 1N4007 diode or something like that would work, with a 1N54xx type definitely covering all the bases as long as it doesnt have to switch on and off faster than about 100Hz. These diodes are pretty cheap even in one piece quantities. The 1N40xx type are fairly rugged too, but the 1N54xx type are even more rugged physically and the leads are heavier so there is less possibility of it breaking off. This is often a concern also for relays and solenoids where there is more than average vibration or other physical shock.