[FLINK-5940] ZooKeeperCompletedCheckpointStore cannot handle broken state handles

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.1.4, 1.2.0, 1.3.0
• Fix Version/s: 1.1.5, 1.2.1, 1.3.0
• Component/s: Runtime / State Backends
• Assignee: Till Rohrmann
• Reporter: Till Rohrmann
• Created: 01/Mar/17 15:43
• Updated: 14/Mar/17 08:27
• Resolved: 14/Mar/17 08:27

Description

The ZooKeeperCompletedCheckpointStore reads a set of RetrievableStateHandles from ZooKeeper upon recovery. It then tries to retrieve the CompletedCheckpoint from the latest state handle. If the retrieve operation fails, then the whole recovery of completed checkpoints fails even though the store might have read older state handles from ZooKeeper.

I propose to harden the behaviour by removing broken state handles and returning the first successfully retrieved CompletedCheckpoint.

Issue Links

• links to: GitHub Pull Request #3446
• links to: GitHub Pull Request #3448
• links to: GitHub Pull Request #3451