News for hackers. Stuff that matters.
Re:Non-blocking transactional atomicity by ch on Tuesday May 28, 2013 @02:54PM (#5781884)
In the mode where a client is responsible to move writes from 'pending' to 'good', it wasn't clear to me what happens if the client dies before contacting all servers in the second pass?

Does the data remain stable, or must some additional work be performed to correct the inconsistent state?

Re:Non-blocking transactional atomicity by pbailis on Tuesday May 28, 2013 @03:20PM (#5782095)
> Does the data remain stable, or must some additional work be performed to correct the inconsistent state?

If you want client writes that reached all servers to become visible, then the servers will have to perform the move from 'pending' to 'good' on their own (by communicating asynchronously). The notification of write stability is idempotent, so it doesn't hurt if both clients and servers perform this notification.

FWIW, in our implementation, servers perform the second step instead of clients (which can be made more efficient via batching).