[FLINK-10354] Savepoints should be counted as retained checkpoints

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.6.0
• Fix Version/s: 1.5.5, 1.6.2, 1.7.0
• Component/s: Runtime / State Backends
• Labels: pull-request-available

Release Note

Savepoints are now used while recovering. Previously when using exactly-once sink one could get into problems with duplicate output data when a failure occured after a savepoint was taken but before the next checkpoint occured.
This results in the fact that savepoints are no longer exclusively under the control of the user. Savepoint should not be moved nor deleted if there was no newer checkpoint or savepoint taken.

• Assignee: Dawid Wysakowicz
• Reporter: Dawid Wysakowicz
• Created: 17/Sep/18 12:35
• Updated: 21/Nov/18 16:47
• Resolved: 21/Nov/18 16:46

Description

This task is about reverting FLINK-6328.

The problem is that you can get incorrect results with exactly-once sinks if there is a failure after taking a savepoint but before taking the next checkpoint because the savepoint will also have manifested side effects to the sink.

Issue Links

• is duplicated by: FLINK-9983 Savepoints should count as checkpoints when recovering
• links to: GitHub Pull Request #6704
• links to: GitHub Pull Request #6858
• links to: GitHub Pull Request #6859