[FLINK-25191] Skip savepoints for recovery

• Type: Sub-task
• Status: Closed
• Priority: Major
• Resolution: Implemented
• Fix Version/s: 1.15.0
• Component/s: Runtime / Checkpointing
• Labels: pull-request-available
• Assignee: Dawid Wysakowicz
• Reporter: Dawid Wysakowicz
• Created: 06/Dec/21 12:17
• Updated: 24/Nov/22 10:54
• Resolved: 16/Dec/21 12:21

Description

Intermediate savepoints should not be used for recovery. In order to achieve that we should:

• do not send notifyCheckpointComplete for intermediate savepoints
• do not add them to CompletedCheckpointStore

Important! Synchronous savepoints (stop-with-savepoint) should still commit side-effects. We need to distinguish them from the intermediate savepoints.

https://cwiki.apache.org/confluence/x/bIyqCw#FLIP193:Snapshotsownership-SkippingSavepointsforRecovery

Document recommendation to drop (change UID) the external state (like transactions from two phase commit sinks) if starting multiple jobs from the same intermediate savepoint.

Issue Links

• relates to: FLINK-29856 Triggering savepoint does not trigger operator notifyCheckpointComplete
• links to: GitHub Pull Request #18092