struggling to find this in the docs; can anyone tell me which longhorn workload pod is responsible for actually communicating with the backup target when creating backups?

I think it might be

longhorn-manager

, but it still depends on what situation you encounter actually. Would you mind to give us more description about the situation?

I’m trying to diagnose why harvester backups aren’t working - it does indeed appear to be handled by longhorn-manager - it seems using a non-routable storage network in harvester wasn’t the smartest idea after all 🙂

longhorn-manager

is responsible for listing all the backups from backup store. If you can’t get

k get lhb,lhbv -n longhorn-system

(backup/backupvolume), you can check longhorn-manager

longhorn-instance-manager-r

is responsible for uploading the volume content to the backup store, so you can also check this component for the uploading issue.