1. When jobs are running, databases may collapse due to restart. Please stop all running jobs when it’s necessary to restart servers.
2. Please do not power off servers directly or use reboot to restart servers. It is recommended to power off servers one after another and then start them one after another. Take three-node cluster as an example (with A as MASTER, B and C as SLAVE):
Power off: node C -> node B -> node A.
Restart: node A -> node B -> node C. Three nodes will be started within 30 minutes.