Hello.
I’m trying to set up an HotSpot Concentrator like in the picture.
I looked in the forum an in the wiki about similar setups but couldn’t find anything helpfull.
I tried with everything bridged (WLAN>EoIP>Bridge Hotspot) but it doesn’t work properly.
I tried with one hotspot service per EoIP, it assigns dhcp ip, asks for login, accepts logins (via radius users) replies the successfull login page, but doesn’t continue to internet (might be ip problem tho). But most of all, the server randomly hangs after login attempt. I have to brutal reboot the machine.
the Server is an x86 with 3.22
Maybe i’m not doing the right setup.
I read in USA MUM there was an example but couldn’t find anything about it in the forum nor wiki.
Browsing a little more the forum i found a lot of users with the same problem of hotspot hanging up with ROS 3.20+ on x86, so i downgraded to 3.13 as someone has fixed his issues.
Still wondering,
Is the picture design correct?
Applying the hotspot server on a bridge with eoip ports in it will work correctly?
Okay seems working with 3.13
The setup is like in the picture, nothing more, nothing less.
Hello, can you go into more detail about this topology? This is exactly the type of solution that we are looking to achieve.
And can I ask you to confirm that you had to revert to an older version of ROS?
I am getting slightly concerned that if we want features working (in your case, complex bridged interfaces, in mine, IPv6 over PPPoE) we have to use an old and unsupported version as newer versions have stopped these features working.
I started this with ROS 3.22 into a VMware virtualmachine.
After some time working, it hangs…
I revert it to 3.13:
/system resource> pr
uptime: 5w1d23h21m56s
version: "3.13"
free-memory: 1799764kB
total-memory: 1824000kB
cpu: "Intel(R)"
cpu-count: 1
cpu-frequency: 2825MHz
cpu-load: 0
free-hdd-space: 461482kB
total-hdd-space: 497357kB
write-sect-since-reboot: 68018
write-sect-total: 68018
architecture-name: "x86"
board-name: "x86"
For remote sites where i cannot have an eoip configured (router behind nat of a dynamic IP of an external ISP) i set up a l2tp bridging and a pp2p to carry a private ip to manage the remote router. Stable and working. Am i going to upgrade it? 
Hi, I’ve set up a test system following the same topology using PPPoE and User Manager on a RB1000 at the backend which in turn connects to the Internet. The AP (RB433AH) between the client (RB433AH) and the AC is just set up with a bridge between the outside and inside facing interfaces. It looks like this…
PPPoE CLIENT (ether1) > AP (ether2)—bridge—AP(ether1) > RB1000 (ether1)—PPPoE server and User Manager—RB1000 (ether4) > Internet.
“ether1” of the RB1000 has no IP address and so without a PPPoE account there is no way of connecting.
This works really well except for SIP which seems to have trouble traversing the NAT on the CLIENT.
The test SIP PBX is connected to the Internet ADSL modem LAN switch.
I find it surprising that this is an issue.
Just to add to the above…
I’ve now set up Hotpsot on the same RB1000 so that clients with a PPPoE profile can auth via User Manager and any client who wishes to prepay can be routed to the Hotspot (which will also be linked to User Manager) and buy credit via Paypal.
Further SIP testing was successful when testing using a remote VoIP account on the Internet so it would indicate the local test SIP PBX is the issue as far as NAT traversal is concerned.
The plan now is to split the backend services and place the Hotspot on its own RB1000 and have PPPoE server and User Manager on the other.