On 04/12/2018 07:46 AM, chen dongming wrote:
> Why did you continue to maintain cassandra part?
I didn't maintain the Cassandra test. It's five years old, written for an
earlier version of Jepsen, and I haven't looked at it since.
> Is it because Cassandra is already bug-free and no need to test?
I very much doubt Cassandra is bug-free, although they have addressed some, and
perhaps all, of the issues I raised in that analysis. There's likely value in
testing it again. I just have other things to work on.
> I still think that Cassandra is not Linearability due to "Last Write Win”
> without lightweight transaction in its latest version, is it
That sounds reasonable. Any LWW system is going to allow lost updates, which
would preclude linearizability.
--Kyle