Hi,
I am going to run 3 Redis sentinels, 2 masters and 3 slaves for each master as separate services on a single stack and same network on docker swarm. Unfortunately, when I remove one of my master services, sentinels can detect that the removed master is down, but they wrongly detect my slaves down as well from the beginning (immediately after they detect the slaves at the beginning, they say that the slaves are down (+sdown)). Therefore, they can not replace the master with a slave and this makes a problem during the failover process. I would deeply appreciate if anyone can help me with this issue.
Kind regards,
Fatemeh