Mailing-List: contact commits-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Date: Fri, 7 Aug 2015 15:49:45 +0000 (UTC)
From: "Stefan Podkowinski (JIRA)" <jira@apache.org>
To: commits@cassandra.apache.org
Message-ID: <JIRA.12852986.1438962570000.6271.1438962585650@Atlassian.JIRA>
In-Reply-To: <JIRA.12852986.1438962570000@Atlassian.JIRA>
References: <JIRA.12852986.1438962570000@Atlassian.JIRA>
 <JIRA.12852986.1438962570371@arcas>
Subject: [jira] [Created] (CASSANDRA-10014) Deletions using clustering keys
 not reflected in MV
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Stefan Podkowinski created CASSANDRA-10014:
----------------------------------------------

             Summary: Deletions using clustering keys not reflected in MV
                 Key: CASSANDRA-10014
                 URL: https://issues.apache.org/jira/browse/CASSANDRA-10014
             Project: Cassandra
          Issue Type: Bug
            Reporter: Stefan Podkowinski


I wrote a test to reproduce an [issue|http://stackoverflow.com/questions/31810841/cassandra-materialized-view-shows-stale-data/31860487] reported on SO and turns out this is easily reproducible. There seems to be a bug preventing deletes to be propagated to MVs in case a clustering key is used. See [here|https://github.com/spodkowinski/cassandra/commit/1c064523c8d8dbee30d46a03a0f58d3be97800dc] for test case (testClusteringKeyTombstone should fail).

It seems {{MaterializedView.updateAffectsView()}} will not consider the delete relevant for the view as {{partition.deletionInfo().isLive()}} will be true during the test. In other test cases isLive will return false, which seems to be the actual problem here. I'm not even sure the root cause is MV specific, but wasn't able to dig much deeper as I'm not familiar with the slightly confusing semantics around DeletionInfo.  


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)