jolshan commented on a change in pull request #11097:
URL: https://github.com/apache/kafka/pull/11097#discussion_r675011109
##########
File path: core/src/test/scala/unit/kafka/server/AbstractFetcherThreadTest.scala
##########
@@ -144,6 +144,44 @@ class AbstractFetcherThreadTest {
assertEquals(2L, replicaState.highWatermark)
}
+ @Test
+ def testDelay(): Unit = {
+ val partition = new TopicPartition("topic", 0)
+
+ class ErrorMockFetcherThread(fetchBackOffMs: Int)
+ extends MockFetcherThread(fetchBackOffMs = fetchBackOffMs) {
+
+ override def fetchFromLeader(fetchRequest: FetchRequest.Builder):
Map[TopicPartition, FetchData] = {
+ throw new UnknownTopicIdException("Topic ID was unknown as expected
for this test")
+ }
+ }
+ val fetcher = new ErrorMockFetcherThread(fetchBackOffMs = 1000)
+
+ fetcher.setReplicaState(partition,
MockFetcherThread.PartitionState(leaderEpoch = 0))
+ fetcher.addPartitions(Map(partition -> initialFetchState(0L, leaderEpoch =
0)))
+
+ val batch = mkBatch(baseOffset = 0L, leaderEpoch = 0,
+ new SimpleRecord("a".getBytes), new SimpleRecord("b".getBytes))
+ val leaderState = MockFetcherThread.PartitionState(Seq(batch), leaderEpoch
= 0, highWatermark = 2L)
+ fetcher.setLeaderState(partition, leaderState)
+
+ // Do work for the first time. This should result in all partitions in
error.
+ val timeBeforeFirst = System.currentTimeMillis()
+ fetcher.doWork()
+ val timeAfterFirst = System.currentTimeMillis()
+ val firstWork = timeAfterFirst - timeBeforeFirst
+
+ // The second doWork will pause for fetchBackOffMs since all partitions
will be delayed
+ val timeBeforeSecond = System.currentTimeMillis()
+ fetcher.doWork()
+ val timeAfterSecond = System.currentTimeMillis()
Review comment:
Ah hmm. It does seem to be a little flaky for the second check
(fetchBackOffMs < secondWorkDuration).
In a sample of 50 tests I ran with backOffMs = 500, there were 8 failures
and it seems like all of them had secondWorkDuration = 500. So maybe I can just
change to <=
Rerunning with this setup 200 times and with fetchBackOffMs=250, I saw 0
failures.
Of course, this was all locally. I'm not sure if Jenkins will behave
differently.
--
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]
