This message was deleted.

I’m wondering if i set “Concurrent Replica Rebuild Per Node Limit” to 0 -> then cordon -> drain -> stop k3s services -> k3kill script -> linux shutdown command would do the trick?

Don't know if setting Concurrent Replica Rebuild Per Node Limit to 0 will avoid rebuilding, maybe you can make a test. You can also have a look here https://github.com/longhorn/longhorn/issues/6921

honestly, i think the most important thing you do before shutting down is gracefully stopping all workloads that use longhorn volumes. either delete them or scale them to 0 replicas i think this is what you'd want to do: 1) scale in longhorn-dependent workloads to 0 replicas 2) cordon nodes 3) drain nodes 4) k3s-killall.sh on all nodes 5) linux shutdown however, presumably all stateful workloads are stopped after 1, so you could likely skip 2 & 3 just remember to scale them out after you bring your cluster back online

Thank you for the feedback.