scaling up and down we’ve ended up with some missing volumes, and we’re struggling to restore them from backup in longhorn, then have the downstream cluster workloads mount/recognize them
specifically I’m really wondering how I can connect a pvc in a downstream cluster to a restored volume in harvester-longhorn
in a longhorn cluster running on the same cluster that consumes longhorn this is straightforward, but not so much when the kubernetes cluster runs in a vm on longhorn, and volumes are mounted in their own hp-volume pod
t
tall-school-18125
08/10/2022, 3:04 PM
@mysterious-boots-59774 can you answer the above question?
m
mysterious-boots-59774
08/10/2022, 3:52 PM
Hi @big-judge-33880, is it in the context of Harvester? You can post your questions in #harvester or #longhorn-storage channel for that.
b
big-judge-33880
08/10/2022, 3:54 PM
I’m not really sure - we use Rancher to provision clusters on top of Harvester, which includes Longhorn, which we consume using the default harvester csi plugin
I believe we were too forceful in draining nodes via rancher, which seems to have caused backing longhorn volumes for our cluster pvcs to disappear
We’ve tried restoring volumes from backup in longhorn directly with original names, created the pv/pvc via longhorn, but we cannot seem to be able to get the downstream cluster running on top of harvester VMs provisioned by rancher to see/use these
I think our main issue may be that the volumemode of restored/longhorn-created PV is Filesystem, not Block, which is the case for the originally provisioned PVs