Omada strange inter-VLAN behavior: Main VLAN can access new VLAN, reverse traffic completely fails
Hi everyone,
I am running into a very strange issue with Omada inter-VLAN routing and I am trying to understand whether this is:
- a bug
- expected behavior
- provisioning corruption
- or something I fundamentally misunderstand.
Setup:
- ER605 v2.0 gateway
- Omada Software Controller (latest v6)
- Omada switch SG2210MP
- VLANs managed directly by Omada
- Dockerized controller running in HOST mode
- No Guest Portal involved for the actual issue
- ACL tests already simplified heavily
Network example:
- Main VLAN: 192.168.14.0/24
- New VLAN: 192.168.21.0/24
Problem:
Devices in VLAN14 CAN access VLAN21.
BUT:
Devices in VLAN21 CANNOT access VLAN14.
Even simple ping fails.
Example:
- 192.168.14.x → 192.168.21.x = works
- 192.168.21.x → 192.168.14.x = fails
What makes this strange:
- The VLAN21 gateway itself (192.168.21.1) is reachable
- Inter-VLAN routing therefore clearly exists
- ACLs were disabled completely for testing
- Isolation settings show ZERO isolated networks
- VLAN21 is NOT configured as guest network
- No client isolation enabled
- Devices tested via both WiFi and Ethernet
- Same behavior reproduced in a second Omada setup at a friend’s house
Important discovery:
I also have a separate WireGuard instance running DIRECTLY on the ER605.
Through THIS WireGuard tunnel, VLAN21 becomes reachable correctly.
This strongly suggests:
- ER605 CAN route the VLANs
- VLAN21 itself is valid
- this is probably not a hardware
Packet captures:
Using packet capture on the Omada switch, I can see ICMP echo requests from VLAN21 devices going to VLAN14 devices, but there is never any response.
Example:
192.168.21.53 → 192.168.14.100 ICMP Echo Request
(no response)
Additional notes:
- I also tested with all LAN→LAN ACLs temporarily set to PERMIT ALL
- Rebooted gateway and switches
- VLANs are directly configured on the ER605 itself
- No static routes should theoretically be required for directly attached VLANs
What confuses me:
This feels too basic to simply be “unsupported”.
Inter-VLAN routing between directly connected VLANs should normally work out of the box.
Has anyone seen:
- asymmetric VLAN communication like this
- stale ACL/provisioning bugs
- Omada controller state corruption
- hidden isolation behavior
- ER605 quirks with software controller setups
Any ideas would be highly appreciated because I am running out of things to test.
Thank You...
