Affected firmware: OpenWrt 23.05.6, Morse-2.11.13. Observed on HaLowLink 1 and HaLowLink 2
Summary: The management-alias daemon /usr/sbin/mgmt-alias-iface hardcodes the same address on every device:
const mgmt_ip = "10.22.121.111";
const dest_network = "10.22.121.110/31";
On startup and on network device link events it adds 10.22.121.111/31 to the active uplink bridge (br-wlan or br-lan). Because the value is identical on all units, any deployment where multiple Morse nodes share an L2 segment — e.g., an 802.11s HaLow mesh bridged to a common LAN — results in every node claiming 10.22.121.111, i.e. a duplicate-IP/ARP conflict.
Symptom: With mesh11sd (morse_mesh11sd 1.2.0) active, this produces continuous Phantom meshnode detected [] - deleting warnings and mesh-path (mpath) flapping between RESOLVED-not-active (0x10) and active (0x15), causing intermittent loss of connectivity to affected nodes and the devices behind them. In a 6-node deployment, one node was flagged 46 times in ~14 h.
Confirmed root cause: stopping/disabling mgmt-alias-iface fleet-wide removed the duplicate address and the phantom churn stopped completely (40+ min clean).
Repro: Bring up two or more HaLow nodes bridging backhaul/AP onto a shared L2 (mesh), with mesh11sd enabled. Observe each node holding 10.22.121.111 and mesh11sd issuing phantom-node deletions.
Suggested fix: Derive the management alias per-node uniquely (e.g., from the device/AL MAC), or allocate a unique /31 per node, or provide a documented config toggle to disable the alias. Note: today the only workaround (/etc/init.d/mgmt-alias-iface disable) is silently undone by firmware updates/factory reset, which re-enable the daemon.
Affected firmware: OpenWrt 23.05.6, Morse-2.11.13. Observed on HaLowLink 1 and HaLowLink 2
Summary: The management-alias daemon /usr/sbin/mgmt-alias-iface hardcodes the same address on every device:
const mgmt_ip = "10.22.121.111";
const dest_network = "10.22.121.110/31";
On startup and on network device link events it adds 10.22.121.111/31 to the active uplink bridge (br-wlan or br-lan). Because the value is identical on all units, any deployment where multiple Morse nodes share an L2 segment — e.g., an 802.11s HaLow mesh bridged to a common LAN — results in every node claiming 10.22.121.111, i.e. a duplicate-IP/ARP conflict.
Symptom: With mesh11sd (morse_mesh11sd 1.2.0) active, this produces continuous Phantom meshnode detected [] - deleting warnings and mesh-path (mpath) flapping between RESOLVED-not-active (0x10) and active (0x15), causing intermittent loss of connectivity to affected nodes and the devices behind them. In a 6-node deployment, one node was flagged 46 times in ~14 h.
Confirmed root cause: stopping/disabling mgmt-alias-iface fleet-wide removed the duplicate address and the phantom churn stopped completely (40+ min clean).
Repro: Bring up two or more HaLow nodes bridging backhaul/AP onto a shared L2 (mesh), with mesh11sd enabled. Observe each node holding 10.22.121.111 and mesh11sd issuing phantom-node deletions.
Suggested fix: Derive the management alias per-node uniquely (e.g., from the device/AL MAC), or allocate a unique /31 per node, or provide a documented config toggle to disable the alias. Note: today the only workaround (/etc/init.d/mgmt-alias-iface disable) is silently undone by firmware updates/factory reset, which re-enable the daemon.