This message was deleted.

Hi James, in this and this documentation I read that for S3 you can configure an URL that is external to the Kubernetes cluster. So I think that if the URL is reachable from your separate network it should go.

Thank you. The S3 bucket won't be accessible from the storage network as it is a separate VLAN / VRF that is not rout-able from the main network. I can just deploy longhorn on the cluster network if needs be. I was just trying to see if the thing that does the backups (not sure if it's the manager, or the engine etc.) performs the backup using that 'storage network' or the main cluster network. Not a problem, if there's no answer in the docs I'll just test it out in our lab. Thanks again.

The docs also say that the storage network is used to transfer data between the engine and replicas, and to read from the backing image, if any. I don't believe that traffic to write to a backup target will use it. Some background info on the purpose and implementation is in the design doc, longhorn/enhancements/20220428-storage-network-through-grpc-proxy.md

Ideal, that's what I was after, thank you.