[CASSANDRA-12867] Batch with multiple conditional updates for the same partition causes AssertionError

• Type: Bug
• Status: Resolved
• Priority: Urgent
• Resolution: Fixed
• Fix Version/s: 3.0.10, 3.10
• Component/s: Legacy/CQL
• Severity: Critical
• Since Version: 3.0.10
• Assignee: Sylvain Lebresne
• Reporter: Kurt Greaves

Authors

Sylvain Lebresne

Reviewers

Benjamin Lerer

• Created: 31/Oct/16 22:09
• Updated: 16/Apr/19 09:30
• Resolved: 07/Nov/16 13:16

Description

Reproduced in 3.0.10 and 3.10. Used to work in 3.0.9 and earlier. Bug was introduced in CASSANDRA-12060.

The following causes an AssertionError:

CREATE KEYSPACE test WITH replication = { 'class' : 'SimpleStrategy', 'replication_factor' : 1 };
create table test.test (id int PRIMARY KEY, val text);
BEGIN BATCH INSERT INTO test.test (id, val) VALUES (999, 'aaa') IF NOT EXISTS; INSERT INTO test.test (id, val) VALUES (999, 'ccc') IF NOT EXISTS; APPLY BATCH ;

Stack trace is as follows:

ERROR [Native-Transport-Requests-2] 2016-10-31 04:16:44,231 Message.java:622 - Unexpected exception during request; channel = [id: 0x176e1c04, L:/127.0.0.1:9042 - R:/127.0.0.1:59743]
java.lang.AssertionError: null
        at org.apache.cassandra.cql3.statements.CQL3CasRequest.setConditionsForRow(CQL3CasRequest.java:138) ~[main/:na]
        at org.apache.cassandra.cql3.statements.CQL3CasRequest.addExistsCondition(CQL3CasRequest.java:104) ~[main/:na]
        at org.apache.cassandra.cql3.statements.CQL3CasRequest.addNotExist(CQL3CasRequest.java:84) ~[main/:na]
        at org.apache.cassandra.cql3.IfNotExistsCondition.addConditionsTo(IfNotExistsCondition.java:28) ~[main/:na]
        at org.apache.cassandra.cql3.statements.ModificationStatement.addConditions(ModificationStatement.java:482) ~[main/:na]
        at org.apache.cassandra.cql3.statements.BatchStatement.makeCasRequest(BatchStatement.java:434) ~[main/:na]
        at org.apache.cassandra.cql3.statements.BatchStatement.executeWithConditions(BatchStatement.java:379) ~[main/:na]
        at org.apache.cassandra.cql3.statements.BatchStatement.execute(BatchStatement.java:358) ~[main/:na]
        at org.apache.cassandra.cql3.statements.BatchStatement.execute(BatchStatement.java:346) ~[main/:na]
        at org.apache.cassandra.cql3.statements.BatchStatement.execute(BatchStatement.java:341) ~[main/:na]
        at org.apache.cassandra.cql3.QueryProcessor.processStatement(QueryProcessor.java:218) ~[main/:na]
        at org.apache.cassandra.cql3.QueryProcessor.process(QueryProcessor.java:249) ~[main/:na]
        at org.apache.cassandra.cql3.QueryProcessor.process(QueryProcessor.java:234) ~[main/:na]
        at org.apache.cassandra.transport.messages.QueryMessage.execute(QueryMessage.java:115) ~[main/:na]
        at org.apache.cassandra.transport.Message$Dispatcher.channelRead0(Message.java:516) [main/:na]
        at org.apache.cassandra.transport.Message$Dispatcher.channelRead0(Message.java:409) [main/:na]
        at io.netty.channel.SimpleChannelInboundHandler.channelRead(SimpleChannelInboundHandler.java:105) [netty-all-4.0.39.Final.jar:4.0.39.Final]
        at io.netty.channel.AbstractChannelHandlerContext.invokeChannelRead(AbstractChannelHandlerContext.java:366) [netty-all-4.0.39.Final.jar:4.0.39.Final]
        at io.netty.channel.AbstractChannelHandlerContext.access$600(AbstractChannelHandlerContext.java:35) [netty-all-4.0.39.Final.jar:4.0.39.Final]
        at io.netty.channel.AbstractChannelHandlerContext$7.run(AbstractChannelHandlerContext.java:357) [netty-all-4.0.39.Final.jar:4.0.39.Final]
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) [na:1.8.0_102]
        at org.apache.cassandra.concurrent.AbstractLocalAwareExecutorService$FutureTask.run(AbstractLocalAwareExecutorService.java:162) [main/:na]
        at org.apache.cassandra.concurrent.SEPWorker.run(SEPWorker.java:109) [main/:na]
        at java.lang.Thread.run(Thread.java:745) [na:1.8.0_102]

The problem is that previous will receive a value after the first statement in the batch is evaluated in BatchStatement.makeCasRequest. I can't see any reason why we have this assertion, it seems to me that it's unnecessary.
Removing it fixes the problem (obviously) but I'm not sure if it breaks something else, or if this is an intended failure case (in which case it should be caught earlier on).

Relevant code is as follows:

CQL3CasRequest.java

    private void setConditionsForRow(Clustering clustering, RowCondition condition)
    {
        if (clustering == Clustering.STATIC_CLUSTERING)
        {
            assert staticConditions == null;
            staticConditions = condition;
        }
        else
        {
            RowCondition previous = conditions.put(clustering, condition);
            assert previous == null;
        }
    }

I've attached a patch that fixes the issue by removing the assert

Issue Links

• is related to: CASSANDRA-12060 Establish consistent distinction between non-existing partition and NULL value for LWTs on static columns