[Question]: OBD server DR handling

[yuing32]

Description

I want to ask about how can I stop/start OCP cluster if my only one OBD server down.

I followed the following official website to setup OCP successfully.
https://www.oceanbase.com/docs/common-ocp-1000000002381076

After that, I want to restart the OCP cluster. So I search about the command. The following command is what I found.
obd cluster stop [cluste_name]
obd cluster start [cluste_name]

Then it come out a problem. If my obd server is out of service, how can I do so?
I tried to deploy the same conf file in another OBD server by following command.
obd cluster deploy [cluste_name] -c conf.yaml
But the cluster status is 'configured', I think it is not a solution.

[longdafeng]

Reject this proposal.

Due to OCP is one devops tools, require a lot of information, it is really hard to do restart and other series command.

At the same time, if you want to stop ocp, you can directly kill ocp process, the problem is how to handle the meta database of ocp. there are a lot of case need to take care.

What's more, most of our user of ocp, they manage a lot of cluster with ocp, most of them wish OCP HA solution, seldom require to stop OCP.

[yuing32]

I think your information focus on OCP, I didn't have any objection about that.
But it is not a reason not to handle OBD's HA/DR function.
If the only OBD server is down, we lose the cluster control function.
Use the worst case, my only OBD server down and my cluster down incidentally. I still can try using second OBD server to start the cluster again if it is existed.
Using other product for sample, like kubernetes cluster, we need three control plane to sure that we still can control pod even one control plane down.