tolbertam commented on code in PR #3917:
URL: https://github.com/apache/cassandra/pull/3917#discussion_r2040858768
##########
test/unit/org/apache/cassandra/cql3/PstmtPersistenceTest.java:
##########
@@ -142,19 +148,103 @@ public void testPstmtInvalidation() throws Throwable
createTable("CREATE TABLE %s (key int primary key, val int)");
+ long initialEvicted = numberOfEvictedStatements();
+
for (int cnt = 1; cnt < 10000; cnt++)
{
prepareStatement("INSERT INTO %s (key, val) VALUES (?, ?) USING
TIMESTAMP " + cnt, clientState);
- if (numberOfEvictedStatements() > 0)
+ if (numberOfEvictedStatements() - initialEvicted > 0)
{
+ assertEquals("Number of statements in table and in cache don't
match", numberOfStatementsInMemory(), numberOfStatementsOnDisk());
+
+ // prepare a more statements to trigger more evictions
+ for (int cnt2 = cnt + 1; cnt2 < cnt + 10; cnt2++)
+ prepareStatement("INSERT INTO %s (key, val) VALUES (?, ?)
USING TIMESTAMP " + cnt2, clientState);
+
+ // each new prepared statement should have caused an eviction
+ assertEquals("eviction count didn't increase by the expected
number", 10, numberOfEvictedStatements() - initialEvicted);
+ assertEquals("Number of statements in memory (expected) and
table (actual) don't match", numberOfStatementsInMemory(),
numberOfStatementsOnDisk());
+
return;
}
}
fail("Prepared statement eviction does not work");
}
+
+
+ @Test
+ public void testAsyncPstmtInvalidation() throws Throwable
Review Comment:
Thanks for the byteman instrumentation suggestion. I added something for
that which enables the test to ensure that all load timestamps precede delete
timestamps and it seems to work quite well.
While it looks like the test is doing quite a bit, it's not actually doing
much more than `testPstmtInvalidation` other than preparing statements
asynchronously with a concurrency of 10 in an executor and with some additional
validation. The test takes on the order of ~1200ms where testPstmtInvalidation
takes ~500ms.
I think there is still some value in it as it is more representative of how
prepared statements can be prepared in practice (multiple threads preparing
statements concurrently), but I'm also ok with just merging the validation i
added here into `testPstmtInvalidation`, let me know what you think :+1:
--
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]
