[FLINK-28964] Release Testing: Verify FLIP-205 Cache in DataStream for Batch Processing

• Type: Sub-task
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 1.16.0
• Component/s: API / DataStream
• Labels: release-testing
• Assignee: jiaxiong
• Reporter: Xuannan Su
• Created: 15/Aug/22 06:15
• Updated: 05/Sep/22 01:53
• Resolved: 31/Aug/22 01:53

Description

DataStream API provides the `cache` method to cache the result of a DataStream and reuse it in later jobs with batch execution mode.

I think we should verify:

1. Follow the doc to write a Flink job that produces cache and a job that consumes cache and submit it to a session cluster(standalone or yarn).
2. You can remove the source physically after the cache-producing job is finished to verify that the cache-consuming job is not reading from the source. For example, delete the file in the filesystem if you are using a file source. 
3. You can restart the TaskManager after the cache-producing job is finished to verify that the cache-consuming job will re-compute the result.