Incorrect Client Detection on Uplink Ports — Omada Controller
Hello TP-Link Omada Team,
I would like to report a client detection issue that affects network monitoring accuracy in environments where Omada-managed switches are connected to a non-Omada core switch.
Environment:
• Managed devices: 5x TP-Link switches (all adopted in Omada SDN)
• Upstream: Non-Omada core switch → Router (DHCP server)
• VLAN configuration: Single VLAN (VLAN 1 / Default), no VLAN segmentation
• LLDP: Enabled on all Omada devices (Gateway, Switch, AP)
Issue #1 — Foreign clients appearing in the Clients section:
Devices that are connected to the core network (upstream of the TP-Link switches) are being displayed in the Omada Clients section as if they were directly connected to one of the managed TP-Link switches. These devices have no physical connection to any Omada-managed switch.
Issue #2 — Client shown on wrong switch / uplink port:
A client device connected to Switch A is being displayed as connected to the uplink port of Switch B. This makes it impossible to accurately determine which switch and port a client is actually connected to.
Root Cause (as I understand it):
Omada appears to rely solely on the MAC address table to determine client location. Since all switches share the same broadcast domain (VLAN 1), MAC addresses of upstream devices are learned on the uplink ports of all managed switches, causing Omada to incorrectly register them as locally connected clients.
UniFi controller in a similar topology (also without VLAN segmentation, also with a non-UniFi core switch) handles this correctly and does not show upstream devices as local clients. This suggests the issue is related to how Omada processes MAC address table entries rather than a fundamental network limitation.
Expected Behavior:
• Only devices physically connected to Omada-managed switch ports (non-uplink) should appear in the Clients section.
• Uplink ports should be excluded from client detection logic.
• Client location should not change between page refreshes.
This issue significantly impacts the usability of Omada SDN for network monitoring in mixed environments (Omada + non-Omada infrastructure), which is a very common real-world deployment scenario.
I would appreciate if this could be reviewed and addressed in a future release.
Thank you.
