As a general rule, micro-interactions are best when they accomplish a single small task very well.
Build a taskflow for your modal and try to simplify it as much as possible. Does the second layer create a new branch in the taskflow? Is that branch necessary? How can you simplify it? What can be removed? Does it have a clear start and a clear end? Is it doing too many things and becoming confusing?
The problem is not so much whether 2 modals is acceptable or not, it's whether the microinteraction itself is clear and simple to accomplish one task well. The discussion of 1 vs 2 modals is just one possible solution.
Test the interface with some users. If not possible, test it with anyone with no knowledge of your product. Points of confusions will likely indicate instances of too much complexity.