[HDDS-4610] Fix issues in 'prepare' operation with one OM down.

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.2.0
• Component/s: Ozone Manager
• Labels: pull-request-available
• Target Version/s: 1.2.0
• Assignee: Aravindan Vijayan
• Reporter: Aravindan Vijayan
• Created: 17/Dec/20 19:13
• Updated: 16/Feb/21 23:41
• Resolved: 16/Feb/21 23:41

Description

On a 3 node OM HA setup, when a prepare is done with 1 OM down, it leads to a state where the leader and follower are fully prepared (Snapshot at last index and logs purged). When the 3rd node rejoins the quorum, it leads to an infinite installSnapshot loop between the leader and the 3rd node, and the system goes into a bad state until a restart is done.

Issue Links

• requires: RATIS-1241 Leader unable to append logs to a recovering follower when its logs have been purged.
• links to: GitHub Pull Request #1836