Hi Nicholas, The issue is the second one you pointed out:
* If they disconnect and reconnect to the wireless, are they assigned the correct VLAN / IP ? This might mean that packetfence is properly associating the new role with the user, but the controller isn't getting dynamically updated. They get the proper IP address….so the issue is when PacketFence needs to update the VLAN via Radius? Still don’t get why the behavior is this one, I’ve checked and the Deauthentication Method is set as RADIUS, Use CoA is enabled, and I even put into the CoA port 1700 since Cisco’s WLC uses that. The only thing that is missing is the Controller IP address field but I don’t think this should cause the issue. Ivan From: Nicholas Pier [mailto:09np...@gmail.com] Sent: lunedì 10 giugno 2019 13:58 To: Ivan Saliu <ivan.sa...@kikocosmetics.com> Cc: packetfence-users@lists.sourceforge.net Subject: Re: [PacketFence-users] Issues with PacketFence Captive Portal configuration Hi Ivan, Let's start with what's supposed to happen immediately after a login: Packetfence should reauthorize the user and send a message to the controller to change the role / vlan. This re-authroization is configured on both the controller and "switch" object in packetfence. Once this comes, the client needs to obtain a new IP address on the new subnet. A few questions then: * Does the client lose network access immediately after the re-authorization? This might indicate a client/controller-side issue if they're not getting a new dhcp lease quickly enough. * If they disconnect and reconnect to the wireless, are they assigned the correct VLAN / IP ? This might mean that packetfence is properly associating the new role with the user, but the controller isn't getting dynamically updated. My "gut" is that this isn't a problem with the way packetfence is deployed (I prefer multiple interfaces, even in VMware), but rather with the controller or "switch" configuration in packetfence. I'd work to verify that SNMP or Radius messages are being sent from packetfence to re-authroize the user and move them to the right VLAN. I think the preferred re-authorization is Radius or CoA since the network configuration guide doesn't mention SNMO read-write configuration. Can you look at the audit tab and verify that packetfence sends a radius message to the controller after login? Similarly, the "switch" config object should have Radius selected from the drop-down for re-authroization. As for the switch, I think it's easiest to not give the switch an SVI IP (vlan interface) and let packetfence do the work. This way you don't need to worry about the routing implications of packetfence having multiple NICs with their own routes, and the ACLs that are required to isolate a client and only allow communication to packetfence. It's just easier. However, if you need to scale to a multi-site deployment and can't tag the registration VLAN end to end, a routed deployment may become necessary. Best wishes, Nicholas P. Pier Network & Virtualization Engineer CCNP RS, PCSNSE7, VCIX6-DCV, VCIX6-NV On Mon, Jun 10, 2019 at 4:51 AM Ivan Saliu <ivan.sa...@kikocosmetics.com<mailto:ivan.sa...@kikocosmetics.com>> wrote: Hi Nicholas, I do agree with you that the flow should be that one. So far I’ve noticed that these points works perfectly: * User connects to SSID and is sent to registration VLAN if their node isn't pre-registered. If the node has been registered, the go immediately to the VLAN associated with their role and this flow stops. * If they're sent to the registation VLAN: * Packetfence provides DHCP and DNS for registration VLAN. * DNS queries from the client are leveraged to redirect them to packetfence for captive portal. Most modern browsers and OSs should do this automatically. After these one though I get to the login page, I log in successfully but I get the error: Unable to detect network connectivity try restarting your web browser or opening a new tab to see if your access has been successfully enabled. After this I’m stuck, the client is not redirected to the new VLAN and I keep the old IP Address. PacketFence is deployed in the following way: 2 NICs, one NIC operates as a Management interface with Radius and DHCP Daemons listening here. The 2nd NIC operates as a Registration Interface, does DNS and DHCP. This is deployed in Hyper-V so this is a forced mode, I cannot use a single interface and build on top VLANs since Hyper-V doesn’t support multiple tagging on this. Also another question: the Registration VLAN, should PacketFence handle the routing? Because right now there is a Cisco 4500 in VSS that is doing the routing. What I’ve also noticed is that the second NIC it is not reachable from outside the subnet but honestly I think this should be how it works since it is supposed to be in an Isolated VLAN. Cannot wrap my head around what I’m missing/what I do wrong. Ivan From: Nicholas Pier [mailto:09np...@gmail.com<mailto:09np...@gmail.com>] Sent: domenica 9 giugno 2019 03:06 To: packetfence-users@lists.sourceforge.net<mailto:packetfence-users@lists.sourceforge.net> Cc: Ivan Saliu <ivan.sa...@kikocosmetics.com<mailto:ivan.sa...@kikocosmetics.com>> Subject: Re: [PacketFence-users] Issues with PacketFence Captive Portal configuration Hi Ivan, I think this is mostly likely a configuration issue. It sounds like you may be expecting the controller to receive information about the captive portal. This may be possible, but it's not how I've deployed packetfence in the past. Instead, Radius and DNS do most of the work I've only worked with 7.x controller code and don't know what's changed since... however the typical workflow I've experienced is as follows: * User connects to SSID and is sent to registration VLAN if their node isn't pre-registered. If the node has been registered, the go immediately to the VLAN associated with their role and this flow stops. * If they're sent to the registation VLAN: * Packetfence provides DHCP and DNS for registration VLAN. * DNS queries from the client are leveraged to redirect them to packetfence for captive portal. Most modern browsers and OSs should do this automatically. * If the user successfully authenticates, packetfence sends a radius message back to the controller to change their VLAN and place them on a different subnet. * Client obtains a new lease and can access the network. I don't know much about your setup, but if its not routed, and clients are placed on the same vlan as packetfence (not a routed deployment). Are you leveraging packetfence for DHCP and DNS on the registration VLAN? Best wishes, Nicholas P. Pier Network & Virtualization Engineer CCNP RS, PCSNSE7, VCIX6-DCV, VCIX6-NV On Sat, Jun 8, 2019 at 7:45 PM Ivan Saliu via PacketFence-users <packetfence-users@lists.sourceforge.net<mailto:packetfence-users@lists.sourceforge.net>> wrote: Hello Guys, I’m experiencing a lot of issues in configuring PacketFence’s Captive Portal (Version 9.0.1) with Cisco’s WLC (5508, software version 8.1). Basically I’ve tried to deploy the solution in two ways: - The “Network Guide” one, where there is only 1 VLAN with ACLs on the WLC to permit only traffic to DHCP/DNS servers and PacketFence Portal. The issue here is the fact that the redirection does not work at all. The Radius parameter with the URL redirection is not filled with data and so the WLC doesn’t redirect at all the traffic. This is an issue because I do not like the user experience, since being force to type an URL to log in and register the device is not good. - The second type of deployment I’ve tried to do is an interface in Registration mode, on a dedicated VLAN managed entirely by PacketFence, trying to use the VLAN change to grant internet access. In this case the Captive Portal works fine, but once I log into it is not recognized internet access and I get an error saying that internet access cannot be validated. If I try to disconnect the client and reconnect it, the VLAN is changed properly and everything works fine, but again this is not a good user experience and I cannot put in a production environment something that doesn’t work properly. This would also be my preferred solution since it grants the best approach to security of course since I would be able to isolate the Registration VLAN and then with Access-List prohibit access to corporate network once the client in registered. Do you have any idea on how to solve these issues? I do think it is most likely a misconfiguration on PacketFence or maybe I’m trying to implement something that it is not supported by Cisco with its WLC?! Any help on this would be greatly appreciated, Ivan _______________________________________________ PacketFence-users mailing list PacketFence-users@lists.sourceforge.net<mailto:PacketFence-users@lists.sourceforge.net> https://lists.sourceforge.net/lists/listinfo/packetfence-users -- Questo messaggio e' stato analizzato con Libra ESVA ed e' risultato non infetto. Clicca qui per segnalarlo come spam.<http://esva.percassi.it/cgi-bin/learn-msg.cgi?id=C717F400D3.A41D3> Clicca qui per metterlo in blacklist<http://esva.percassi.it/cgi-bin/learn-msg.cgi?blacklist=1&id=C717F400D3.A41D3> -- Questo messaggio e' stato analizzato con Libra ESVA ed e' risultato non infetto. Clicca qui per segnalarlo come spam.<http://esva.percassi.it/cgi-bin/learn-msg.cgi?id=EBE41400CB.ACCF2> Clicca qui per metterlo in blacklist<http://esva.percassi.it/cgi-bin/learn-msg.cgi?blacklist=1&id=EBE41400CB.ACCF2>
_______________________________________________ PacketFence-users mailing list PacketFence-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/packetfence-users
