[CASSANDRA-15683] Fix NPE in SimpleReadWriteTest after test framework behavior change

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 4.0, 4.0-alpha4
• Component/s: Test/dtest/java
• Bug Category: Correctness - Test Failure
• Severity: Low
• Complexity: Low Hanging Fruit
• Discovered By: Unit Test
• Platform: All
• Since Version: 4.0-alpha

Source Control Link

https://github.com/apache/cassandra/commit/6ae65967a5ed789164192104a3e220980f952e1c

Test and Documentation Plan

no doc changes

• Assignee: Kevin Gallardo
• Reporter: Kevin Gallardo

Authors

Kevin Gallardo

Reviewers

Brandon Williams, David Capwell

• Created: 02/Apr/20 15:44
• Updated: 21/Dec/20 08:08
• Resolved: 02/Apr/20 21:11

Description

In JVM dtests, all exceptions thrown in a request execution used to be always wrapped in a RuntimeException.

After CASSANDRA-15650 changes, the behavior has been changed to: If the exception is a RuntimeException, just rethrow it, otherwise wrap in RuntimeException, as you can see in: https://github.com/apache/cassandra/commit/dfc279a22a5563ac7a832a586914d5410426e9b7#diff-0b019281b7e97248577c82af0e663ef4R211

This causes the tests that were always extracting the cause from the wrapping RuntimeException before, to check the root cause of the error, to throw a NPE when they call getCause(), tests such as SimpleReadWriteTest#readWithSchemaDisagreement and SimpleReadWriteTest#writeWithSchemaDisagreement.

Can be fixed by simply not unwrapping the cause in those tests, use the thrown exception directly, if the behavior of "not always wrapping in RuntimeException" is agreed to be correct.

Issue Links

• is related to: CASSANDRA-15650 Fix flaky test org.apache.cassandra.distributed.test.*RepairCoordinatorFastTest
• is related to: CASSANDRA-15539 Extract in-jvm API and tests out of Cassandra and into a separate repository