adelapena commented on code in PR #2369:
URL: https://github.com/apache/cassandra/pull/2369#discussion_r1210553787


##########
test/distributed/org/apache/cassandra/distributed/test/guardrails/GuardrailPartitionSizeTest.java:
##########
@@ -75,12 +75,11 @@ protected Cluster getCluster()
     @Test
     public void testPartitionSize()
     {
+        // test yaml-loaded config
         testPartitionSize(WARN_THRESHOLD, FAIL_THRESHOLD);
-    }
+        schemaChange("DROP TABLE %s");
 

Review Comment:
   The tests extending `o.a.c.distributed.test.guardrails.GuardrailTester` use 
a single cluster for the entire suite. The regular test case uses the yaml 
configuration set at startup, the one that is set at `setupCluster`. In 
contrast, the test for dynamic update changes that config on the running 
instance by simulating the JMX methods.
   
   That works if the test for static config runs before the test for dynamic 
update, but not the other way around. If the test for dynamic config runs 
first, the test for static config will find the dynamically updated config, 
breaking the test. We could simply restart the node before each test, but I 
think the test will save resources if we simply combine both tests. I think the 
tests are brief enough to remain readable.
   
   So `GuardrailPartitionSizeTest` only worked incidentally because the tests 
were run in the fortunate order. The new `GuardrailPartitionTombstonesTest` 
however was not so fortunate, exposing the bug. So I have changed both to make 
them independent of whatever order the test runner decides for us.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to