[
https://issues.apache.org/jira/browse/CASSANDRA-7902?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14351267#comment-14351267
]
Aleksey Yeschenko commented on CASSANDRA-7902:
----------------------------------------------
The primary reason for going LOCAL_* CL for reads is to avoid latency hit, so I
still think that crossing local dc for RR there is at least somewhat surprising
to some. Not limiting RR to local for LOCAL_* CL was an oversight - I don't
remember the issue being raised at the time at all.
Now, swapping the default global/local RR chances in CASSANDRA-7320 was
definitely the right thing to do, but I believe it's still orthogonal to this
ticket. I don't have much data either, but tying RR candidate sourcing to
per-request CL vs. just the per-table config does feel more precise. It *is*
reasonable to sometimes use different consistency levels on the same table
based on the data being queried. That I know, because I've done it before - and
we still do a form of that for auth, switching CL based on whether or not we
are querying for the default su.
I will not fight strongly for it, but I really want to make this change in 3.0.
Partially because it's less surprising behavior, partially because it will
simplify code a bit - but mostly the former. It would also allow us to resolve
this ticket without any extra options/levels.
Any other opinions? [~thobbs] [~jbellis] [~rssvihla]?
> Introduce the ability to ignore RR based on consistencfy
> --------------------------------------------------------
>
> Key: CASSANDRA-7902
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7902
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Brandon Williams
>
> There exists a case for LOCAL_* consistency levels where you really want them
> *local only*. This implies that you don't ever want to do cross-dc RR, but
> do for other levels.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
