Our small Linstor cluster is connected via a single 25G switch. Occasionally, we need to reboot this switch or perform other maintenance on the network that makes it unavailable for short periods. We’d like to setup a backup link over another 1G switch to be able to bridge this period, ideally failing over automatically and moving back to the fast link when it comes online again.
I’ve been pouring over the documentation but I haven’t yet been able to achieve this.
I’ve created the second interface on all nodes and set the 25G interface as PrefNic:
root@node-a:~# linstor node interface list node-a
╭──────────────────────────────────────────────────────────────╮
┊ node-a ┊ NetInterface ┊ IP ┊ Port ┊ EncryptionType ┊
╞══════════════════════════════════════════════════════════════╡
┊ + ┊ backup_nic ┊ 10.77.4.2 ┊ ┊ ┊
┊ + StltCon ┊ default ┊ 10.77.3.2 ┊ 3366 ┊ PLAIN ┊
╰──────────────────────────────────────────────────────────────╯
root@node-a:~# linstor node lp node-a
╭──────────────────────────────────────────────╮
┊ Key ┊ Value ┊
╞══════════════════════════════════════════════╡
┊ Aux/NodeName ┊ node-a ┊
┊ CurStltConnName ┊ default ┊
┊ DrbdOptions/AutoEvictAllowEviction ┊ false ┊
┊ NodeUname ┊ node-a ┊
┊ PrefNic ┊ default ┊
╰──────────────────────────────────────────────╯
With this configuration, Linstor does not fall back to backup_nic when default goes offline.
What else needs to be configured? How would I fail over manually? The documentation suggests manually creating paths, but that is not practical with hundreds or thousands of resources.
Thanks in advance.