[FLINK-21400] Attempt numbers are not maintained across restarts

• Type: Sub-task
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 1.13.0
• Component/s: Runtime / Coordination
• Labels: pull-request-available
• Assignee: Chesnay Schepler
• Reporter: Chesnay Schepler
• Created: 18/Feb/21 18:08
• Updated: 25/Mar/24 16:10
• Resolved: 12/Mar/21 10:33

Description

The DeclarativeScheduler discards the ExecutionGraph on each restart attempt, as a result of which the attempt number remains 0.

Various tests use the attempt number to determine whether an exception should be thrown, and thus continue to throw exceptions on each restart.

Affected tests:
UnalignedCheckpointTestBase
UnalignedCheckpointITCase
ProcessingTimeWindowCheckpointingITCase
LocalRecoveryITCase
EventTimeWindowCheckpointingITCase
EventTimeAllWindowCheckpointingITCase
FileSinkITBase#testFileSink

Issue Links

• causes: FLINK-34409 Increase test coverage for AdaptiveScheduler
• links to: GitHub Pull Request #15019
• links to: GitHub Pull Request #24284