[IOTDB-620] [Distributed] Add the ability to persist and maintain voteFor for the raft node and amend the judgment conditions for elections

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Done
• Component/s: Core/Cluster
• Labels: distributed
• Assignee: Tan xinyu
• Reporter: Tan xinyu
• Created: 27/Apr/20 09:01
• Updated: 20/May/20 03:59
• Resolved: 29/Apr/20 15:59

Description

This is the quote from raft's paper.

Raft servers must persist enough information to stable storage to survive server restarts safely. In particular, each server persists its current term and vote; this is necessary to prevent the server from voting twice in the same term or replacing log entries from a newer leader with those from a deposed leader. Each server also persists new log entries before they are counted towards the entries’ commitment; this prevents committed entries from being lost or “uncommitted” when servers restart.

Therefore, the persistence of voteFor is important to ensure the security of the system.

In addition, some judgment logic of the election module is inconsistent with that described in the paper, which may lead to bugs in some corner cases. Therefore, it is very important to correct the implementation method described in the paper.

It will be the work of this issue to implement the voteFor persistence and fix some election logic.

Issue Links

• is a child of: IOTDB-68 IoTDB supports distributed version