Hacker News Reader

The problem here is that the request timeout is greater than the lock timeout.

While this might make this situation more likely to occur, you can never prevent concurrent accesses from happening in a distributed system.