No, they explain why in the article: Their monitoring was looking at the application (nginx), so packets getting stuck in the kernel weren't reflected in that.