One day, on the faraway world of not-medieval-Europe, someone discovered that a type of stone could be made to float. A decade or two later, there are flying castles, built out of and/or on a foundation of said floating stone.
Floating stone doesn't actually float everywhere. There are large lines (let's call them Ley lines) crisscrossing continents and oceans, where stone floats best on, separated by voids where it doesn't float much, or even at all. Altitude above ground also weaken the effect, so you won't see floating castles more than a few km up at most.
If a floating castle drifts away from a Line, it will (hopefully) gently float down and land - though "gently" at the scale of a castle can still be potentially fatal to anything both in or under the castle.
What propulsion would be used on a floating castle with medieval technology?
Standard airship techniques may not work, as flying castles are so incredibly denser and more massive. Ideally, it would allow for a fair bit of control, to avoid getting pushed out of a Line or into a mountain.
Available tech can be from late Roman to mid-late medieval era. There is no magic apart from floating rock, and it doesn't really scale down.
On the other hand, there have been a few decades with incentives to solve new problems, so new tech can be used if it could have reasonably been invented in those conditions. Likewise, flying castles themselves can help in engineering projects, for example to (slowly) move very large charges up.
The following numbers and assumptions are given as a baseline, but can be changed if it makes things more interesting:
The two main things moving a drifting floating castle are wind and natural currents from the Ley line. Generally, the Ley current is weaker though not negligible, and constant on one place. A castle can also very slowly gain and loose altitude. Assuming no ballast drop, a castle can accelerate at about 5 cm/s², with a max ascending speed of 2m/s, in good conditions (descending speed is limited only by lithobraking).
The largest flying foundations are roughly 30m radius circles, with 20m-thick foundation. The smallest ones are 8m radius and 6m-thick foundation. The foundation itself is at least 9/10 filled in stone, but there are often small chambres and tunnels for access below. There can be light structures hanging from under the foundation, though most of the mass must be over it.
Floating stone has a density of about 2.5 g/cm³. So the biggest castles have a foundation of about 150,000 tons, and the smallest of 3,000 tons. (If other properties are needed, we can use granite as placeholder)
Total non-floating-stone carrying capacity of a castle is about 1/3 of the stone mass, for the best-built (and most expensive) models - so 1,000 to 50,000 tons.
To save on capacity, walls can be partially built with floating stone, up to about 1/3 of the mass of the foundation. Those walls must be at least 1m thick, but don't count on carrying mass and can be used as support for other structures.
Flying castles typically have a smaller slab inside, in a long vertical shaft, capable of independent vertical movement. It is usable only when there is little to no horizontal acceleration, and with a max vertical speed of 10 cm/s. It can be used to raise about 1/30 of the mass of the foundation (so 100 to 5,000 tons). Some people are experimenting with cisterns at both ends, using hydraulic power from the upper cistern, then raising spent water from the lower to the upper one. Not sure how well this scheme would work - this would be the subject of another question, but consider that there is some power available from this, if it can help.