[
https://issues.apache.org/jira/browse/CASSANDRA-8272?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16042891#comment-16042891
]
Andrés de la Peña commented on CASSANDRA-8272:
----------------------------------------------
It seems that there were some thrift-related dtests failing in the path for
3.11. While fixing them I have realized that Thrift commands send only the
fetched columns, so they may not send the queried columns that are required to
apply the index filter in the coordinator side. The indexed column value are
actually fetched but filtered
[here|https://github.com/apache/cassandra/blob/cassandra-3.11/src/java/org/apache/cassandra/index/internal/keys/KeysSearcher.java#L182-L193]
since [CASSANDRA-11523|https://issues.apache.org/jira/browse/CASSANDRA-11523].
So we could just move this filter to the coordinator-side, probably to
[ReadCommand#postReconciliationProcessing|https://github.com/adelapena/cassandra/blob/8272-3.11/src/java/org/apache/cassandra/db/ReadCommand.java#L447-L464].
However, I think that if we do such replica-side change we would end having the
same problem with upgrades that prevents us to apply the full solution to 3.x.
That is, not-upgraded replicas could send rows without the
indexed-but-not-fetched columns to an upgraded coordinator that would reject
them. Complementary, upgraded replicas could send rows including the
indexed-but-not-fetched columns to not-upgraded coordinators that would return
them without applying the row filter. Probably we could also have problematic
scenarios during reconciliation.
Here is a fixed patch that just skips coordinator-side filtering of index
results for Thrift commands:
||[3.11|https://github.com/apache/cassandra/compare/cassandra-3.11...adelapena:82b122b1ce5b172e11b4be7f02fdb7581bd28291]|[utests|http://cassci.datastax.com/view/Dev/view/adelapena/job/adelapena-8272-3.11-testall/]|[dtests|http://cassci.datastax.com/view/Dev/view/adelapena/job/adelapena-8272-3.11-dtest/]|
||[trunk|https://github.com/apache/cassandra/compare/trunk...adelapena:1416d9b082d7f93b187cbf67abd9a917735c4804]|[utests|http://cassci.datastax.com/view/Dev/view/adelapena/job/adelapena-8272-trunk-testall/]|[dtests|http://cassci.datastax.com/view/Dev/view/adelapena/job/adelapena-8272-trunk-dtest/]|
This means that not included index implementations couldn't benefit from our
consistency fix when using Thrift. Included index implementations weren't going
to do so anyway because we are not going to apply the replica side of the fix
in 3.x.
No news for trunk.
What do you think? Is it acceptable to don't apply the changes to Thrift
commands?
> 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: Andrés de la Peña
> Fix For: 3.0.x
>
>
> When replica return 2ndary index results, it's possible for a single replica
> 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 are
> done at {{QUORUM}}:
> {noformat}
> UPDATE test SET v = 'bar' WHERE k = 0;
> SELECT * FROM test WHERE v = '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 (since
> 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), instead of
> skipping that tombstone, we'd insert in the result a corresponding range
> tombstone.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
