Linstor satellite using undesired interface

[jonathon2nd]

We have and existing linstor cluster on baremetal, and we want to use linstor csi for k8s clusters.

For single interface clusters, using the storage vlan only for testing, we have no issues. However when we go to deploy a prod like cluster (eth0 public, eth1 private, eth2 storage only), the chart/operator wants to use eth1.

I see no way to modify this in helm values. Is this something that could be added, or be able to use node labels/annotations to get the desired ip

https://github.com/piraeusdatastore/piraeus-operator/blob/64c3d13c4d22b52c4a2a5e0075dbf4d8de288bb1/pkg/controller/linstorsatelliteset/linstorsatelliteset_controller.go#L549 I see no way to overwrite this.

[WanzenBug]

Would it be an option to use eth1 for linstor-internal communicationm and eth2 only for DRBD replication? If so, you can just set up a secondary interface on the node (linstor node interface create <nodename> <ifname> <ip>), and set the PrefNic property on the nodes: linstor node set-property PrefNic <ifname>.

Otherwise, there is no option to configure this currently.