While Jepsen (and this article) is focused on behavior under node failure and network partitions, this caught my eye: > It also exhibits Stale Read, Lost Updat

While Jepsen (and this article) is focused on behavior under node failure and network partitions, this caught my eye: > It also exhibits Stale Read, Lost Update, and other forms of G-single in healthy clusters This looks like quite a fundamental issue.