We usually take 3-4 high-priority, rather big features into sprint. We estimate them in story points. During the sprint we also fix some urgent, not related to these features bugs. We estimate those as soon as they are dropped into sprint as well to measure "real velocity". Note, that those bugs are never in a sprint goal.
I'm thinking if it's a good practice. Or should we drop it and just measure velocity based on these high-priority features being done? Any thoughts?