[KUDU-422] Replicated master process (no master SPOF)

• Type: New Feature
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: M5
• Fix Version/s: 1.0.0
• Component/s: consensus, master
• Labels: roadmap-candidate
• Target Version/s: 1.4.0
• Assignee: Adar Dembo
• Reporter: Todd Lipcon
• Created: 31/Jul/14 12:25
• Updated: 18/Jun/19 13:02
• Resolved: 10/Mar/17 20:09

Description

Support for multiple masters with consensus-replicated data. Goal is to have to be able to do manual failovers if the leader of the master quorum dies, stretch goal for this milestone is to have that failover happen unattended.

Sub-Tasks

• Replicated master process: tests for SysTable replication.: Resolved
• Replicate master process: manual reconfiguration and failover for master quorums: Resolved
• Replicate master process: make client aware of multiple masters: Resolved
• Replicated master process: implement SysTables replication: Resolved
• Unattended fail over using failure detection and leader election: Resolved
• Master should handle the case where we REPLICATE fails: Resolved
• Following a master leader election, create table may fail: Open
• Master does not rebuild replica information soft state on failover: Resolved
• Operations triggered by TS heartbeats may go unperformed: Resolved
• Provide a mechanism to migrate from single to multiple masters: Resolved
• Web UI and metrics support for replicated master: Resolved
• Changes to SysTables should be queryable on follower leaders upon replication: Open
• Replicate master process: send heartbeats to all masters.: Resolved