What is actor reentrancy and why is it dangerous?
The essence
An actor is reentrant: when one of its methods reaches an await and suspends, the actor is not blocked — it can start executing another call (including the same method from another task). When the first call resumes, the actor's state may no longer be what it saw before the await. A data race is still impossible (only one task runs inside the actor at a time), but invariants break: double loading, double charging, a stale cache. That is a race condition without a data race.
The rest of this answer, with code and interview follow-ups, is in the DevPrep app.
Read the full answer in DevPrep