This message was deleted.

That's a good question. Ultimately the block copy has to come from longhorn and egress to the backup target. I would guess that unless your backup target is on the storage network it's going to use the cluster gateway to traverse out the management network. (I don't really know though it's just what makes sense to me network wise)

Yes IIUC, so the backup reads the volume data from replicas/snapshots which uses storage network if configured, but backup sending it to the NFS/S3 is external which is via the mgmt interface with default routing. Currently we do not use additional or secondary network for external traffic and its not implemented yet. cc: @magnificent-pencil-261

During backup, the volume data is sent through the management network. During restore, the volume data is restored through the management network and then forwarded to the Storage Network for each individual replica restoration. In other words, that part of the data traffic is internal to LH. As @nice-horse-9611 mentioned, Harvester does not support using a secondary interface for backup and restore.

Hello, Thank you for the quick response. I appreciate your help!