First part of #1644: C*3 from 3.11.11->3.11.12

[tatu-at-datastax]

What this PR does:

Starting work on #1644: first upgrade Cassandra-3: separate upgrades to isolate problems.

Instructions to follow:

• https://github.com/stargate/stargate/tree/master/persistence-cassandra-3.11#cassandra-version-update

Which issue(s) this PR fixes:

Will eventually resolve #1644 but not with just this PR.

Checklist

• Changes manually tested
• Automated Tests added/updated -- relies on test coverage
• Documentation added/updated
• CLA Signed: DataStax CLA

[tatu-at-datastax]

Alas, immediate (unit) test failures for seemingly trivial update:

Step #4 - "cassandra-3.11": [INFO] Running io.stargate.db.cassandra.Cassandra311PersistenceActivatorTest
Step #4 - "cassandra-3.11": [INFO] Tests run: 2, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.48 s - in io.stargate.db.cassandra.Cassandra311PersistenceActivatorTest
Step #4 - "cassandra-3.11": [INFO] 
Step #4 - "cassandra-3.11": [INFO] Results:
Step #4 - "cassandra-3.11": [INFO] 
Step #4 - "cassandra-3.11": [ERROR] Errors: 
Step #4 - "cassandra-3.11": [ERROR]   StargateQueryHandlerTest.authorizeByTokenBatchStatement:936 ? IllegalState Can...
Step #4 - "cassandra-3.11": [ERROR]   IdempotencyAnalyzerTest.shouldReturnIdempotentIfAllStatementsWithinABatchAreIdempotent:102 ? IllegalState
Step #4 - "cassandra-3.11": [ERROR]   IdempotencyAnalyzerTest.shouldReturnNonIdempotentIfAllStatementsWithinABatchAreNonIdempotent:117 ? IllegalState
Step #4 - "cassandra-3.11": [INFO] 

[ivansenic]

Hey @tatu-at-datastax, can you please go through the list that I made with the last update -> https://github.com/stargate/stargate/tree/master/persistence-cassandra-3.11#cassandra-version-update and check all the points.. Could be that thrift version moved, and is not back in the transitive deps of the 3.11..

Also the list is missing the docker files updates for the v2.. Maybe we can add a PR for the v2 branch once everything is updated on the master.

[ivansenic]

Stops from merging, due to the #1647 (comment). Same issue exists in the 3.11.12 as well.

[ivansenic]

The other failing tests here are reproducible locally..

Step #4 - "cassandra-3.11": [ERROR] Errors: 
Step #4 - "cassandra-3.11": [ERROR]   StargateQueryHandlerTest.authorizeByTokenBatchStatement:936 ? IllegalState Can...
Step #4 - "cassandra-3.11": [ERROR]   IdempotencyAnalyzerTest.shouldReturnIdempotentIfAllStatementsWithinABatchAreIdempotent:102 ? IllegalState
Step #4 - "cassandra-3.11": [ERROR]   IdempotencyAnalyzerTest.shouldReturnNonIdempotentIfAllStatementsWithinABatchAreNonIdempotent:117 ? IllegalState

Seems like a bug due to this changed line in BatchStatement.Parsed. The isFullyQualified() is called, which throws the IllegalArgumentException in the super class CFStatement. I don't really understand how can a batch be fully qualified, this has to be a bug.. This should be confirmed by the BatchStatement.Parsed constructor that calls super with null.

This bug does not exist in the 4.0.2 as they correctly override the isFullyQualified() method.

[tatu-at-datastax]

As per comment on parent issue, some issue(s) could be due to:

https://issues.apache.org/jira/browse/CASSANDRA-17248

which changes behavior wrt fully-qualified prepared statements. Patch is sizable, against 3.0 it's apache/cassandra@242f7f9 (merged upwards)

[tatu-at-datastax]

Completed work from update list, but as to the real problem(s), I am not quite sure how to proceed.

[ivansenic]

Seems that calling QueryProcessor.getStatement instead of QueryProcessor.parseStatement does the trick and batches are correctly prepared in that case.. If tests are green, I will update all the tests existing in all persistence modules to not use QueryProcessor.parseStatement..

Seems we only use this in tests, so should be fine overall..

[ivansenic]

Hmm, OK tests are green, but getStatement does not provide all options as the parseStatement, so I opt in leaving everything else as is. Would this be OK?

There is still the race condition issue here as well, not sure if #1647 would fix that. If so, this should be merged after.

[ivansenic]

@tatu-at-datastax Approving, but please make sure race condition is fixed in #1647

[tatu-at-datastax]

Still need to resolve the race condition (see #1647) which does also affect C*3.11, before releasing this PR.