Everything Rene said is correct, but I'd like to add a few things. In general: the more physics you add to the model, the better things will get in terms of avoiding "unphysical" behavior.
If all parts of the model (Fluid properties in Medium, and state equations) support two-phase, the liquid will start to boil, increase in volume, and in that way avoid the negative pressure. This process can be fast and violent and cause numerical difficulties with low tolerances.
In reality, and good models for cavitation, dissolved gases in the liquid bubble out before boiling in many relevant situations (hydraulic oils). That makes the onset of cavitation less violent, and is a common way of modeling cavitation in hydraulic circuits. Good hydraulics libraries, like the one from Modelon, can (if cavitation is enabled) capture these effects.
In other cases: when you hit 0 pressure, you are always outside the region of validity of the model.