[
https://issues.apache.org/jira/browse/FLINK-20655?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17260257#comment-17260257
]
Qingsheng Ren commented on FLINK-20655:
---------------------------------------
I used Flink integrated state machine example with 4 parallelisms for testing,
and a topic with 32 partitions on Kafka clusters. These scenarios are tested:
# Unbounded mode running for ~1 hour given a fixed topic
# Use topic pattern to consume from multiple topics
# Use topic pattern and create a new topic for testing new topic discovery
# Scale out a topic for testing new partition discovery
# Unbounded mode with a stopping offset
# Bounded mode
# Kill a TM then restart
# Stop the job with a snapshot and resume
All these cases looks good to me.
> Add E2E tests to the new KafkaSource based on FLIP-27.
> ------------------------------------------------------
>
> Key: FLINK-20655
> URL: https://issues.apache.org/jira/browse/FLINK-20655
> Project: Flink
> Issue Type: Test
> Components: Connectors / Kafka
> Affects Versions: 1.12.0
> Reporter: Jiangjie Qin
> Assignee: Qingsheng Ren
> Priority: Major
> Fix For: 1.13.0, 1.12.1
>
>
> Add the following e2e tests for KafkaSource based on FLIP-27.
> # A basic read test which reads from a Kafka topic.
> # Stop the job with savepoint and resume.
> # Kill a TM and verify the failover works fine.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
