Describe the bug
redis version:6.0.8
A cluster with 3 sentinels, 1 master, 1replica. When master down , cluster failover, the replica change to be the new master. after 5 mins later, stop the new master and restart the old master after new master odown. The old master change to a new master again.
A short description of the bug.
Failover caused data loss.
To reproduce
redis version:6.0.8
1. stop the master node A redis service
2. wait for failover , the node B redis will change to a new master.
3. set key to the master
4. stop the node B redis (new master), wait for odown
5. start node A redis, the node A redis will change to a new master again, but node A redis will loss data that writed to B redis
Steps to reproduce the behavior and/or a minimal code sample.
Expected behavior node A redis start agian after node B redis stop and odown, the node A redis cannot change to a new master A description of what you expected to happen.
Additional information sentinel key params: down-after-milliseconds 30000 failover-timeout 180000
Any additional information that is relevant to the problem.
Comment From: watchpoints
wait for failover , the node B redis will change to a new master. start a