EAP783_V1_1.1.92 Beta Firmware -- Closed

EAP783_V1_1.1.92 Beta Firmware -- Closed

142 Reply
Re:EAP783_V1_1.1.90 Beta Firmware (Released on 17th Jan, 2025)
Yesterday

  @TW_EPC 

 

 

> Now, however, with the 1.1.92 beta, a problem has arisen in VLAN 1 where the connections are no longer clean. Is there anything known about this? In addition to VLAN 1, the customer also uses many other VLANs in the network where there are no problems.

 

 

I can confirm that this behavior is still present in the current "stable" firmware:
EAP783 1.0_1.1.4 Build 20251030.
 

My setup was fairly standard:
- Default LAN on untagged VLAN 1 (so, effectively "no VLAN" for LAN/Wi-Fi clients)
- Multiple additional VLANs for things like IoT
- Routing and firewalling handled centrally, nothing exotic


The symptoms were extremely frustrating and non-obvious:

- Traffic from default VLAN 1 Wi-Fi to other VLANs was unreliable, with what looked like missing TCP ACKs / asymmetric connectivity.
- Concrete example: I could not reliably access Valetudo on a robot vacuum that lives on its own IoT VLAN, when connecting from Wi-Fi on default VLAN 1.
- The same access worked perfectly from wired LAN in the same default network, so the problem clearly sat in the EAP/VLAN handling on the wireless side, not in routing or firewall rules.


After way too much packet-tracing and trial-and-error, the only stable workaround I’ve found is:

- Stop using VLAN 1 / untagged as the "real" LAN.
- Create a dedicated "LAN" VLAN for normal devices (tagged on the trunk, access on the appropriate ports/SSIDs).
- Move all "normal" LAN clients (including the Wi-Fi SSID that used to sit on default VLAN 1) to this new non-1 VLAN.

Once I did that, inter-VLAN communication immediately became reliable again. No more broken sessions, no more partial loads, no more weird one-way traffic.

I’m shocked by how much time I’ve burned debugging because of the EAP783. The rest of the VLAN design is straightforward, and every other VLAN behaves correctly, only VLAN 1 on the EAP783 is problematic.


@Vincent-TP 

Is this a known regression on EAP783 (1.1.92 beta and 1.0_1.1.4 Build 20251030)?

And more importantly, is there a plan to fix VLAN 1 behavior on Wi-Fi so we don’t have to design around it by avoiding VLAN 1 entirely?

Right now, the practical advice I’d give anyone deploying EAP783 is: do not use VLAN 1 as your production LAN, put your "real" LAN on a dedicated tagged VLAN and avoid VLAN 1 wherever possible.

  0  
  0  
#143
Options
Re:EAP783_V1_1.1.90 Beta Firmware (Released on 17th Jan, 2025)
Yesterday

Hi  @nikisima 

 

Thanks for the feedback. Please send a copy of your controller's config file to the ticket TKID251228355. I just created it for you.

This will help us find the reason as soon as possible. Thanks.

nikisima wrote

  @TW_EPC 

 

 

> Now, however, with the 1.1.92 beta, a problem has arisen in VLAN 1 where the connections are no longer clean. Is there anything known about this? In addition to VLAN 1, the customer also uses many other VLANs in the network where there are no problems.

 

 

I can confirm that this behavior is still present in the current "stable" firmware:
EAP783 1.0_1.1.4 Build 20251030.
 

My setup was fairly standard:
- Default LAN on untagged VLAN 1 (so, effectively "no VLAN" for LAN/Wi-Fi clients)
- Multiple additional VLANs for things like IoT
- Routing and firewalling handled centrally, nothing exotic


The symptoms were extremely frustrating and non-obvious:

- Traffic from default VLAN 1 Wi-Fi to other VLANs was unreliable, with what looked like missing TCP ACKs / asymmetric connectivity.
- Concrete example: I could not reliably access Valetudo on a robot vacuum that lives on its own IoT VLAN, when connecting from Wi-Fi on default VLAN 1.
- The same access worked perfectly from wired LAN in the same default network, so the problem clearly sat in the EAP/VLAN handling on the wireless side, not in routing or firewall rules.


After way too much packet-tracing and trial-and-error, the only stable workaround I’ve found is:

- Stop using VLAN 1 / untagged as the "real" LAN.
- Create a dedicated "LAN" VLAN for normal devices (tagged on the trunk, access on the appropriate ports/SSIDs).
- Move all "normal" LAN clients (including the Wi-Fi SSID that used to sit on default VLAN 1) to this new non-1 VLAN.

Once I did that, inter-VLAN communication immediately became reliable again. No more broken sessions, no more partial loads, no more weird one-way traffic.

I’m shocked by how much time I’ve burned debugging because of the EAP783. The rest of the VLAN design is straightforward, and every other VLAN behaves correctly, only VLAN 1 on the EAP783 is problematic.


@Vincent-TP 

Is this a known regression on EAP783 (1.1.92 beta and 1.0_1.1.4 Build 20251030)?

And more importantly, is there a plan to fix VLAN 1 behavior on Wi-Fi so we don’t have to design around it by avoiding VLAN 1 entirely?

Right now, the practical advice I’d give anyone deploying EAP783 is: do not use VLAN 1 as your production LAN, put your "real" LAN on a dedicated tagged VLAN and avoid VLAN 1 wherever possible.

 

  0  
  0  
#144
Options