Overview

This document explains TSC Split brain and Split brain Solution.

Method

1. TSC network and storage configuration

1) TSC network

2) TSC storage configuration

2. TSC Split brain

1) TSC Split brain CASE

 - If NODE #2 is accidentally started as FAILOVER in a normal environment, the VIP of NODE #2 comes UP and clients connect to NODE #2.

 - All existing connections to the VIP of NODE #1 are disconnected, and new connections are made to the VIP of NODE #2.

 - When started by user error on NODE #2, it basically starts in MOUNT mode.

NODE2$ tbboot
Change core dump dir to /tsc/tibero6/bin/prof.
Listener port = 8629
********************************************************
* Critical Warning : Raise svmode failed. The reason is
* TBR-1098 : Database in standby mode. Failover is required to boot as primary.
* Current server mode is MOUNT.
********************************************************
Tibero 6
TmaxData Corporation Copyright (c) 2008-. All rights reserved.
Tibero instance started suspended at MOUNT mode.

- When the user starts NODE #2 as FAILOVER, the VIP of NODE #2 comes UP, causing the VIP of NODE #1 to not function properly and connections are made to the VIP of NODE #2.

- If data comes in through the VIP of NODE #2, data mixing issues may occur, so you must choose whether to use NODE #1 or NODE #2.

2) TSC Split brain Solution

•  AFO: Auto Failover 

- When the conditions for AFO occur, the PRIMARY can no longer fulfill its role, and the TARGET node provides service as FAILOVER.

- If the conditions for AFO are not met, the TARGET node does not fail over.

• If the conditions for performing AFO are met, the PRIMARY node is unable to provide normal service

  Type	PROTECTIVE	DISASTER_PLAN
  1) CM abnormal	X	O (AFO)
  2) CM down	X	X
  3) DB abnormal	O (AFO)	O (AFO)
  4) DB down	O (AFO)	O (AFO)
  5) network failure	X	O (AFO)
  6) cmfile failure	X	O (AFO)