Hi Jonathon,

It does work very very well and you are on the right track but there are a few 
gotchas to watch for.

On each router on your network you will need to create a registration VLAN.

DO NOT set up IP HELPERS for the VLANs as this will not work properly. Remember 
that DHCP ACKS are the final stage in a complete DHCP allocation
process and since DHCP-ACKs are unicast and not broadcast packets the IP helper 
will not receive them. Your PF server will be blind to them and will miss
clients coming and going on the network. Without this you will get the 'Your 
device is not found on the Packetfence database...' message a lot!

You will also need to enable 802.1X on your wired interfaces and/or WiFi SSIDs 
to use 'Mac Authentication'. This is how the device gets captured by the
PF server's 'captive portal' it looks at the MAC address sent and ALWAYS 
returns an 'Access-Accept' reply. But it will also return other RADIUS 
attributes
such as dynamic VLAN assignment for the 'Registered/Unregistered' status of the 
node.

On your PF server:-

Each remote routed VLAN will need it's associated subnet to be defined as a 
'registration' network.

Configure your PF server to only act on DHCP-ACKs - it can see any that it is 
the DHCP server for but will never see the remote ones (see UDP reflector 
later).

Configure the network switches on your PF server to apply 'Role based VLAN 
assignment' and assign VLANs for each role you have.

Each registration subnet must be defined to use the registration interface for 
its default gateway (ie the interface that the captive portal will use
to talk to the client device). IPTables on the PF sever will manage and forward 
access to/from this subnet to the captive portal process for you.

You also need to add in the subnets of all the networks that the clients will 
be moved to (via role based VLAN assignment). These subnets
need to be defined (on your PF server) to use the management interface of your 
PF server as its default gateway.

On your main production DHCP server (not the PF server):-

Define a scope for each registration subnet that specifies that the subnet use 
your PF's server registration interface's IP address as its sole DNS server.
Also make sure you have a shortish lease time (say 30 minutes) as this will 
help with enabling the 'Parking' feature' that despite the documentation to the
contrary, you can use despite the PF server not being the DHCP server.

Set up a 'UDP reflector' service that duplicates all the DHCP traffic and sends 
to the management interface of your PF server (it uses a form of encapsulation).
The pfdhcplistener on the PF server will process the DHCP traffic and populate 
the nodes and IP database for you as devices come and go.
Remember here that, until PacketFencer listen to me and sort out the 
pfdhcplistener subroutines, your PF server will populate your nodes database
with absolutely everything that uses your DHCP server (servers/ IP phones/WiFI 
AP's etc).
I myself have custom tweaked the code (/lib/pf/dhcp/processor.pm) on my PF 
server to eliminate all but the subnets that are defined on the PF server
in order to keep the nodes database to a manageable size.

You can and should leave the DHCP option 'enabled' for all the subnets that you 
define for registration and client deployment (The PF server will not see the
DHCP requests for the remote subnets since you are not using IP Helpers).

It takes a bit of playing around but once you have it set up it becomes a lot 
clearer, if you need to see some configs then please PM me.

Andrew Torry



-----------------------------
     Falmouth University
-----------------------------

-----Original Message-----
From: Jonathan Hornby via PacketFence-users 
[mailto:[email protected]]
Sent: 05 September 2017 10:28
To: [email protected]
Cc: Jonathan Hornby; Steven Swart
Subject: Re: [PacketFence-users] DHCP forwarding required?

Good day

I am trying to setup packetfence to act as a captive portal on a routed 
network. I have many different end user sites which all sit on my routed 
service provider network. I want to run the packetfence server centrally in 
conjunction with my Ruckus virtual SmartZone controller as the wireless and 
captive portal solution.

My problem is that Packetfence seems to want DHCP requests either tunneled to 
itself, or forwards by means of a DHCP relay, so that Packetfence itself can be 
the DHCP server to all devices which will authenticate to it. I really need the 
DHCP function to remain on each respective local DHCP server at each site.

So my question is, is packetfence just wanting to know the MAC address related 
to each IP address for internal authentication purposes? Or does packet fence 
actually require itself to be the DHCP server? Should the former be true, how 
do I set it up to work this was without Packetfence acutally being the DHCP 
server?
OR, if the latter is true, then how can I set Packetfence DHCP server up to 
know which pool/gateway/DHCP options to offer based on the SSID which the 
device came through, as I need to offer devices the correct DHCP settings on 
their own subnet based on the site that they are at.

Please let me know if anything is unclear so that I can clarify.

Thank you
Jonathan


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
PacketFence-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/packetfence-users
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
PacketFence-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/packetfence-users

Reply via email to