So I hate to admit I need a tiny bit of assistance. I've exhausted my very shallow understanding of these few things. I guess big picture, we deployed Tenable Security Center via Helm Chart. It installed, we licensed it, and used it for like a month. The Node it was on rebooted for an underlying OS update and it refused to start. Re-installed the Helm Chart, but now its bound to another PVC. Is there a way for me to re-bind it to the previous? Note: I did actually backup the volume in Longhorn (not not internal to the application) to an external S3 target.

I think you can either restore your backup and bind it that way (if it's in a new namespace)

I'm learning how little I know, and somehow...how I mamaged to get this far. I think the Helm chart puts everything in the "tenable" namespace on creation. The backups attached the old PVC are in that namespace too, but I'm not finding any option to restore to another PVC from that backup. The Old PVC is actually up but "released" in Longhorn, on the same node as the Pod. Is there some magic I am missing where I could pull a swap? Or get the Pod to come up attached to this PVC?

You need to edit the PVC to point at the Old PV not the new one that it's attached to.

So turns out, I believe the issue was you were right, but something was locking the mount point on the node. I drained that node, and rebooted (actually had a pending OS update anyways)...and it came back on another node without issue.

Probably just needed to kill the pod after editing the PVC

Yeah, so much learning for me this week...Pretty sure the pod was on when the original node rebooted, so probably never killed the pod correctly.

Typically specially when using helm charts, the pods are kinda nested under a different object that defines them. Typically a DaemonSet or Deployment or StatefulSet. Typically this is where you go to change a setting or object. It's also where you can re-deploy the pods that exist. It's useful for when there's health checks because it'll wait until the new pod is up and running before killing the old one keeping your service up while you replace the pods.

StatefulSet in this case. I am pretty out of touch on some level since I was using the Rancher UI...some of it isnt' obvious by me

image.png

I wrote a little script yesterday to make sure I cordon/drain properly in the future before a reboot. That might have been a lion's share of the problem. Longhorn is just toooo good at replicas. lol