How much feedback to give for an undergraduate / master's thesis?
As noted in a comment, this is probably highly dependent on the university culture and field. My answer is based on my impressions from CS in Germany.
In my subfield (HCI-related), it was most usual for doctoral candidates to supervise between one and four Bachelor/Master theses a year, maybe a bit less for post-docs (each of which took exactly 6 months). (Depending on the topic, supervision occasionally took place in groups - i.e. two or three supervisors for one student, taking turns at supervision tasks or focusing different aspects of the work.)
As a rough guideline, the "amount" of supervision provided was usually a weekly meeting (typically, between 10 and 60 minutes in length) with the student to give them advice and answer questions. And, realistically, in some phases and for some students, to remind them that they should get some work done because they cannot work on their Bachelor/Master thesis forever. As a secondary channel of supervision, students were free to ask urgent questions whenever they came up, via e-mail or by just dropping by in the supervisor's office whenever it was convenient.
The kind of supervision covered a wide spectrum - essentially, everything from high-level advice (decisions about the content and how to proceed, developing some basic ideas for solutions together, scientific best practices, ...) to low-level operational details (technical issues in the code and in the LaTeX document, stylistic writing and layouting questions in the document, ...) could come up.1
As for looking at drafts, I agree with semi-extrinsic's answer in that it's likely too late if the thesis is almost due. Personally, when I knew the student was in the phase in which they had already scheduled some writing, I would ask my students on almost every meeting whether there's a new or revised section that I can read and comment on (not necessarily because I was so keen on commenting, but first and foremost as another means of urging students to do the part of the task that most liked the least - writing the thesis document). Moreover, at least at my home department, we usually required students to use the department's Subversion version control system, which was another way how supervisors could look at the students' progress and the current state of their thesis when they had a feeling progress wasn't as smooth as it should be.
Lastly, concerning fairness, I must admit the thought that there could be anything unfair (in a way that should be avoided) in this never crossed my mind. Of course, supervisors in different institutes may hold very different opinions as to what amount and kind of effort is optimal and feasible in their supervision. But then, students are not assigned to supervisors in my place; they usually pick a topic they like and they know beforehand who is going to be the supervisor for that topic. Anyway, as a result of the different opinions, supervision between institutes differed. Students could have bad luck with their supervisor, or with the typical style of supervision in the institute that focuses on the subfield they like. But the question is: Should you do anything about it? I'd say no. First, because I consider it part of a university graduate's skills to also get along with supervision that does not ideally suit their personal preferences. And moreover, because I firmly believe that unfairness should not be countered by removing any advantages until every option is as bad as the original quality minimum.
1: This does not mean that students would have supervisors do all their work. The goal was rather guiding the students towards a balance between working out problems autonomously and not being shy to actively contact the supervisor about any current roadblocks.
As their primary supervisor,I would expect you to give them as much feedback as you can (whithin reason).
I also think that getting a draft when their theses are soon due is too late.
As to your comment that masters students should be more independent than PhD students - wouldn't you expect the opposite?
One other note that I would add to the existing good answers: the quality of your students work reflects on you as well.
I would never even consider graduating a student whose thesis I had not carefully reviewed and given sufficient feedback on to ensure that it was not embarrassing to have my name associated with. This is lower stakes for an undergraduate or Masters thesis than it is for a Ph.D. thesis, because there is less expected of the student.
Nevertheless, one of your important jobs as an supervisor is to ensure that sufficient standards of academic quality are upheld. In practice, this will generally require some degree of feedback and mentoring in order to get the students to a level of quality that you are comfortable with.