Hi All, I am seeking clarification regarding a bare metal use case involving OpenStack 25.02 and OVN 25.03. Specifically, I would like to understand how Bug 1995078 <https://bugs.launchpad.net/neutron/+bug/1995078>(“OVN: HA chassis group priority is different than gateway chassis priority”) applies to my current topology and what the best practices are.
Environment Details: - Neutron: ML2/OVN - Logical Router: One router with an External VLAN network (for SNAT/DNAT/FIPs) and an Internal VLAN tenant network hosting bare metal servers (plus Geneve tenant networks for VMs). - Gateway Chassis: Present, but no HA chassis groups are configured yet. Currently, a single active gateway chassis handles all external traffic. - reside-on-redirect-chassis has been set to true, but the traffic from compute node is not hitting ovn-gateway chassis In a deployment with one effective OVN gateway chassis and no HA chassis groups, is Bug 1995078 (or related bugs 2092271/2052821) expected to affect bare metal external ports on VLAN tenant networks? If I maintain a single gateway chassis without HA, are there recommended OVN/Neutron configurations—such as ovn-chassis-mac-mappings, router LRPs, or ha_chassis_group usage—to avoid intermittent connectivity or ARP issues when mixing VLAN bare metal and Geneve VM networks? When introducing HA chassis groups in the future, is there a documented pattern to ensure the active chassis for bare metal external ports and the chassis handling the router’s external/internal VLAN LRPs remain in sync without manual priority realignment? I can provide the OVN Northbound configuration if it would be helpful for your analysis. Thank you for your guidance and for the work already done to mitigate these issues. Best regards, Srini
_______________________________________________ discuss mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
