Ok, now I have storage questions... I've read thro...
# harvester
s
Ok, now I have storage questions... I've read through all the RAID discussion threads in this channel. One thing still confuses me. If I have a multi-server/node Harvester cluster, even if I use hardware RAID on the server, I still need to ensure I have a replica of that data on another server's storage if I want the pod attached (KubeVirt VM in this case) to said volume to be able to be rescheduled anywhere else. So, even if I use the hardware RAID (to improve read/write performance on single server), I'm still going to need to incur the Longhorn CPU/Network overhead to manage creating the replicas on 2 other servers, right? While this isn't technically necessary for a drive failure (because of the RAID), it's necessary for Kubernetes pod scheduling as the servers cannot access the local drive storage of any other server. I CAN see where this may improve performance (CPU) on single server/node Harvester cluster because I can disable Longhorn entirely. Let me know if I'm missing something. Thanks!