Best Practices for Longhorn Node Scheduling

Hello all,

I have been setting up Longho/?rn through Helm Charts and using persistence volumes and storage classes.

I have started down the journey of wanting to have certain nodes to be dedicated ‘longhorn storage servers’ and not have any other workloads on them except for the longhorn components.

Is it just a matter of applying a taint with Noschedule option?
Have you gone down this path?

Many thanks,

You can use arbitrary taints with NoSchedule or NoExecute effect for those dedicated nodes then add the corresponding tolerations for Longhorn.

You can check the following doc for more details: