Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Date: Mon, 15 May 2017 12:35:04 +0000 (UTC)
From: "Sylvain Lebresne (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12753516.1415308953000.210177.1494851704497@Atlassian.JIRA>
In-Reply-To: <JIRA.12753516.1415308953000@Atlassian.JIRA>
References: <JIRA.12753516.1415308953000@Atlassian.JIRA> <JIRA.12753516.1415308953192@jira-lw-us.apache.org>
Subject: [jira] [Commented] (CASSANDRA-8272) 2ndary indexes can return stale
 data
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
archived-at: Mon, 15 May 2017 12:35:10 -0000


    [ https://issues.apache.org/jira/browse/CASSANDRA-8272?page=3Dcom.atlas=
sian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D=
16010420#comment-16010420 ]=20

Sylvain Lebresne commented on CASSANDRA-8272:
---------------------------------------------

Looks good to me.

bq. for trunk, we could either remove the {{Index#customExpressionFor(CFMet=
aData cfm, ByteBuffer value)}} implementation or keep it as an ease for ind=
ex implementations not interested in or able of implementing the coordinato=
r side part of this.=20

For what it's worth, I'd really remove it. The default we're providing is j=
ust broken for any sensible index implementation, and have a default implem=
entation that is intrinsically broken is wrong. Sure, so far we haven't rel=
ied on that implementation, and with this ticket we would rely on it in a p=
articular case only, but all this isn't very future proof. And there is no =
real cost to asking custom index to implement it: worst case, they can just=
 making it always return true themselves if they so wish, but at least they=
 do it while being plainly aware that it's dodgy and could brake things in =
the future.

> 2ndary indexes can return stale data
> ------------------------------------
>
>                 Key: CASSANDRA-8272
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8272
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Sylvain Lebresne
>            Assignee: Andr=C3=A9s de la Pe=C3=B1a
>             Fix For: 3.0.x
>
>
> When replica return 2ndary index results, it's possible for a single repl=
ica to return a stale result and that result will be sent back to the user,=
 potentially failing the CL contract.
> For instance, consider 3 replicas A, B and C, and the following situation=
:
> {noformat}
> CREATE TABLE test (k int PRIMARY KEY, v text);
> CREATE INDEX ON test(v);
> INSERT INTO test(k, v) VALUES (0, 'foo');
> {noformat}
> with every replica up to date. Now, suppose that the following queries ar=
e done at {{QUORUM}}:
> {noformat}
> UPDATE test SET v =3D 'bar' WHERE k =3D 0;
> SELECT * FROM test WHERE v =3D 'foo';
> {noformat}
> then, if A and B acknowledge the insert but C respond to the read before =
having applied the insert, then the now stale result will be returned (sinc=
e C will return it and A or B will return nothing).
> A potential solution would be that when we read a tombstone in the index =
(and provided we make the index inherit the gcGrace of it's parent CF), ins=
tead of skipping that tombstone, we'd insert in the result a corresponding =
range tombstone. =20


--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org