cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sergio Bossa (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-5062) Support CAS
Date Wed, 27 Feb 2013 16:51:14 GMT


Sergio Bossa commented on CASSANDRA-5062:


{quote}if the leader reports success to the client after local commit, but before it has been
committed to the followers, we could either (1) lose the commit on failover if followers are
pessimistic, or (2) commit data that we originally reported failed{quote}

Nope, once the commit is reported as successful, even if still not acked, it will be always
seen by clients because:
1) If the leader fails, the failed over one is guaranteed to see up to the latest (proposed
or committed) value, because that's the way it is elected.
2) The only case when a commit can fail happens when the prepare phase doesn't get a quorum,
but in this case the leader will have to retry until it gets it, times out or fails; last
two options do not mean to the client the commit has failed, just that it has to retry (I
think the same would be with Paxos, as the client is never part of the consensus).

This is really Zab/ZK by the way, I'm not adding much to it :)
> Support CAS
> -----------
>                 Key: CASSANDRA-5062
>                 URL:
>             Project: Cassandra
>          Issue Type: New Feature
>          Components: API, Core
>            Reporter: Jonathan Ellis
>             Fix For: 2.0
> "Strong" consistency is not enough to prevent race conditions.  The classic example is
user account creation: we want to ensure usernames are unique, so we only want to signal account
creation success if nobody else has created the account yet.  But naive read-then-write allows
clients to race and both think they have a green light to create.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message