|
From: Erik <er...@va...> - 2020-04-23 17:07:26
|
On 23-04-2020 18:19, Ludovic Zammit wrote: > Hello Erik, > > If you check the routed network documentation you can see an example > for a remote site. > > https://packetfence.org/doc/PacketFence_Installation_Guide.html#_routed_networks > > With VLAN enforcement you would need to have one registration network > - VLAN per remote site. > > On that remote registration VLAN interface you would configure an IP > helper toward your PacketFence layer2 registration interface. Once you > create that, On PacketFence you create the remote registration network > and PacketFence would know which IP to distribute based on the network. I had seen that, thanks. But that seems to imply that PF must be configured with each individual network. And I want to avoid that. We are talking about many hundreds of sites/networks here. All of this is handled by the VPN system already. Each site is provisioned via a web portal, where the IP range is defined and sent to the sites DHCP-server. I could add a module to the VPN system, that sends information about each site to PF. But the DHCP service must remain on site. If only to prevent problems should a site be unable to contact PF. > > You would also need to create a switch configuration on PacketFence to > authorize the radius authentication incoming from that remote switch. > DHCP and Radius are two separate workflow. Exactly. And I want to keep them separate. AAA by PF. And DHCP locally. I don't actually have use case for profiling yet, but does it actually require PF to be the DHCP server. Or can it do profiling if a local DHCP helper somehow informs PF of which IP was locally assigned to which client? I guess I will have to look into Fingerbank to see how that works in detail. |