Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: tiup 重启tidb 节点
I previously thought that using tiup to restart TiDB server nodes would restart them one by one in sequence. However, today I tried it in a production environment with 3 TiDB server nodes, and the restart order was that 2 TiDB server nodes were restarted first, followed by the third TiDB server node. It seems that they are not restarted one by one in sequence. You can try it out yourselves. I’m not sure if my conclusion is correct.
What version is the issue related to?
As long as the node’s service capability can be maintained, it’s fine. As for how to restart the node, you can still do it manually.
When the first 2 TiDB nodes restart, is the service unavailable?
This issue has never occurred in v5.4.0 after a restart.
There is another node available.
I was just a bit curious, I thought it was restarting one by one before.
As long as they don’t all restart at the same time 
Is it not required for the majority to be online?
I haven’t paid attention to this, but in a production environment, it is still recommended to manually specify a node for the restart to ensure safety and uniqueness.
In theory, it is a rolling upgrade by polling.
For TiDB nodes, the majority is not necessary; as long as one is online, it’s fine.
Thank you for your reply.
Indeed, both TiDB and PD are stateless and do not require majority consensus.
The TiDB server is stateless, so as long as you inform the application side in advance that there will be a brief interruption, you can restart it boldly.
This topic was automatically closed 1 minute after the last reply. No new replies are allowed.
|
username: Raymond
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: