Honestly? Most of them, at least as far as the European Middle Ages are concerned. We hear about the big splashy successful sieges, but that's mostly because siege warfare was almost always a military failure for the attacking side. The logistics and realities of a siege heavily favored defendants in most cases. Fortresses (castles or fortified parts of towns) stocked up on food supplies, but often were able to trade even with enemy soldiers looking to make some quick money for extra food. And it wasn't just the defenders who had to feed themselves! The besieging army had to eat off the surrounding land, probably filled with hostile people, and certainly less likely to be planted again once plundered.

Furthermore, castle defenses were generally pretty good and frequently improved. Cologne in 1206 was indeed besieged and sacked. What was the immediate reaction? Rebuild the defenses and make them even stronger, so it wouldn't happen again. Even the introduction of gunpowder weaponry to a defensive system established for hurtled projectiles was no guarantee. It took 1514 large cannonballs to break down Maastricht's defenses in a lengthy campaign over 1407-08; in 1480, besiegers of Rhodes expended 3800 cannonballs in their attempt. Defenders learned to build second walls, to build earthwork supports behind stone walls. In fact, apparently a lot of victories over castles and towns in the early gunpowder years came about through intimidation rather than actual strength! The only "pregnable" thing was people's emotions.

Again, this is not to say that cities and castles never fell, and definitely not to undercut the utter devastation wrought by sackers and conquerors. However, numerically speaking, being the offensive party in a siege campaign was far more often a losing proposition on the military field. Its primary power was through intimidation and coercion to negotiate--a power that the occasional victory certainly did very much to reinforce.