Jimmy Wang created KAFKA-18638:
----------------------------------
Summary: Flasky test
ClientIdQuotaTest.testThrottledProducerConsumer
Key: KAFKA-18638
URL: https://issues.apache.org/jira/browse/KAFKA-18638
Project: Kafka
Issue Type: Bug
Components: core
Affects Versions: 4.0.0
Reporter: Jimmy Wang
Assignee: Jimmy Wang
Attachments: ScreenShot_20250124154202.PNG
[https://ge.apache.org/scans/tests?search.rootProjectNames=kafka&search.tags=trunk&search.timeZoneId=Asia%2FShanghai&tests.container=kafka.api.ClientIdQuotaTest&tests.sortField=FLAKY&tests.test=testThrottledProducerConsumer(String%2C%20String)%5B2%5D]
Also flasky for testQuotaOverrideDelete.
This could happen when numRecords is set to 1000. Even if there is no quota
validation, the function quotaTestClients.produceUntilThrottled(numRecords)
would also return because it has already produced 1000 messages.
When numRecords is decreased from 1000 to 500, this case could occur more
frequently in my local environment.
The value of defaultProducerQuota is 8000 bytes, and the size of each message
is about 6 bytes. This means that we should produce at least 160 messages per
second to reach the rate limit.
I think what we should do to fix it is:
1. Increase the message size.
2. Decrease the defaultProducerQuota.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
