[
https://issues.apache.org/jira/browse/CASSANDRA-5138?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13549808#comment-13549808
]
Sylvain Lebresne commented on CASSANDRA-5138:
---------------------------------------------
I don't think it is reasonable to have CQL validate on every read that no bad
data has been added to non compact table through thrift: outside of not being
particularly simple to implement and error prone, it would be silly performance
wise.
The fact is, *non* compact CQL3 tables are a CQL3 things and in theory they
should not be accessible at all through thrift (exactly because thrift has the
potential of screwing things up). But I do understand the willingness to use
thrift to look under the cover to see how things are implemented and that
definitively have educational value. Which imo leaves us with the following
options:
# do nothing and be clear that 'you shalt not mess up with non compact CQL3
table through thrift' and that if you do, things may break in unexpected ways.
# add validation on the thrift write path.
# refuse write access to non compact CQL3 tables from thrift by default with an
option to deactivate that protection "at your own risk".
In an ideal world 2) would be the best solution. However, adding said
validation is not 2 lines of code, far from it, and there would be the
maintenance cost of keeping said validation up to date with the evolutions of
CQL3. So, and especially given that accessing non compact CQL3 table from
thrift is difficult anyway due to not all metadata being exposed, I'm not
convinced that adding said validation would be the best use of our developer
resources. Of course if someone is willing to step up to write and maintain
that validation code, that's fine, but in the absence of that I would suggest
3) as a good enough compromise.
> Provide a better CQL error when table data does not conform to CQL metadata.
> ----------------------------------------------------------------------------
>
> Key: CASSANDRA-5138
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5138
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.2.0
> Environment: Mac OS X running 1.2
> Reporter: Brian ONeill
> Priority: Minor
> Attachments: northpole.cql
>
>
> When you create a table via CQL, then insert into it via Thrift. If you
> inadvertently leave out a component of the column name, in CQL you receive a:
> TSocket read 0 bytes
> Server-side the following exception is logged:
> ERROR 15:19:18,016 Error occurred during processing of message.
> java.lang.ArrayIndexOutOfBoundsException: 3
> at
> org.apache.cassandra.cql3.statements.ColumnGroupMap.add(ColumnGroupMap.java:43)
> at
> org.apache.cassandra.cql3.statements.ColumnGroupMap.access$200(ColumnGroupMap.java:31)
> at
> org.apache.cassandra.cql3.statements.ColumnGroupMap$Builder.add(ColumnGroupMap.java:138)
> at
> org.apache.cassandra.cql3.statements.SelectStatement.process(SelectStatement.java:805)
> at
> org.apache.cassandra.cql3.statements.SelectStatement.processResults(SelectStatement.java:145)
> at
> org.apache.cassandra.cql3.statements.SelectStatement.execute(SelectStatement.java:134)
> at
> org.apache.cassandra.cql3.statements.SelectStatement.execute(SelectStatement.java:61)
> at
> org.apache.cassandra.cql3.QueryProcessor.processStatement(QueryProcessor.java:132)
> at
> org.apache.cassandra.cql3.QueryProcessor.process(QueryProcessor.java:140)
> at
> org.apache.cassandra.thrift.CassandraServer.execute_cql3_query(CassandraServer.java:1686)
> at
> org.apache.cassandra.thrift.Cassandra$Processor$execute_cql3_query.getResult(Cassandra.java:4074)
> at
> org.apache.cassandra.thrift.Cassandra$Processor$execute_cql3_query.getResult(Cassandra.java:4062)
> at org.apache.thrift.ProcessFunction.process(ProcessFunction.java:32)
> at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:34)
> at
> org.apache.cassandra.thrift.CustomTThreadPoolServer$WorkerProcess.run(CustomTThreadPoolServer.java:199)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
> at java.lang.Thread.run(Thread.java:680)
> I'll submit a schema, and steps to reproduce.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
