I TA'd a first course in abstract algebra during my senior year of undergrad. The professor wanted a computational flavor to the course, so we introduced Magma right off the bat. We wanted to allow the students to experiment with permutations without needing to actually do large computations themselves. As part of my contribution to the course, I was put in charge of the Magma side of things, both making the assignments and teaching them how to use the software.
After a few in class lectures on Magma syntax and a couple ungraded, basic-syntax mini-assignments, I gave the students this, which they finished in about two weeks. I had to help them a lot by email. Everyone finished the second problem once they figured out the list syntax. In retrospect I think the first problem was a bit too difficult. I had expected the students to have at least a little programming experience, and it turned out most of them didn't, so the Monte Carlo type solution was not intuitive. They just weren't comfortable enough with conditionals and control structures to use them to tackle unfamiliar concepts.
I gave two more small projects like this, but I don't have copies online. The second one involved verifying the Chinese remainder theorem, calculating isomorphism classes of abelian groups of some given orders, and taking quotients of abelian groups. The third one was about alternating groups and conjugacy classes.
It turned out that the students found Magma to be more trouble than it was worth. They didn't learn as much from the assignments as I had expected them to. The goal had been to allow students to see more complex group structure in action without actually having to go through the trouble of calculating everything on their own, but it seems that this difficulty was simply replaced by the difficulty of learning the new language. I had expected they would enter an "exploratory phase" after learning the basic tools where they used learned a whole lot of group theory (because that's what happened to me after I learned Magma), but it didn't happen.
So, my experience suggests that even several days in the computer lab was not sufficient to introduce Magma. I suspect it may only be practical to use Magma for a first course in group theory when you know that students have at least some previous programming experience. If they do, I still think it could be a great way to learn. It also might be worth considering use in a first year graduate level course in abstract algebra, where the students are committed enough to put in the time to learn programming concepts if they don't know how to yet. (Also, at least for me, the most educational part of learning Magma was messing around with the SmallGroups package, which is probably out of the scope of a first year course in group theory.)
P.S., to answer your final question, the only resource I have ever personally used to learn Magma has been the Magma handbook, which has been so tremendously helpful that I have felt the need to seek an alternative.
EDIT: I remembered that I had made some Magma notes for my students, a quick reference and some short programs we went over in class.