LAN access from VPN
-
I assigned the interface (WG_VPN) to the tunnel, so I already have both rules.
-
@ghazel So the outbound NAT should be automatically created then.
Do you see the tunnel network listed? Firewall/NAT/OutboundMake sure the MTU is at 1420 on the wireguard interface.
Diagnostics/Routes, tunnel listed and on correct interface?
-
@jarhead I don't see anything interface specific in Firewall/NAT/Outbound, but I do see the 192.168.2.0/24 range.
Similarly for Diagnostics/Routes, attached screenshot of what seems relevant:
-
@ghazel Post pics of rules and configs from both ends.
-
@jarhead not certain which rules and configs you're talking about. One side is pfSense, which I've shown above, the other end is a macOS machine running wireguard. So no rules there, and only the most basic wireguard config.
-
@ghazel When you say "other" LAN devices, are you actually trying other devices or just one? It might be a software firewall on that device if just one.
-
@jarhead Any other device on 192.168.1.1/24 is unreachable. No firewalls on those devices. I guess I could packet capture to see what happens to those packets..
-
Packet capture on the pfSense LAN interface does show the packets from 192.168.2.14 (WG_VPN subnet) destined for 192.168.1.16 (LAN subnet). However, there is no response. Perhaps tellingly, 192.168.1.16 can reach 192.168.2.14 and it does respond. So for example I can ssh from 192.168.1.16 to 192.168.2.14, but not from 192.168.2.14 to 192.168.1.16.
-
@ghazel but there is a rule on the VPN interface allowing access to lan?
Usually it’s:
Missing fw rule in arriving interface
Device fw doesn’t allow different subnet
Device not using pfSense as gateway -
@ghazel said in LAN access from VPN:
Packet capture on the pfSense LAN interface does show the packets from 192.168.2.14 (WG_VPN subnet) destined for 192.168.1.16 (LAN subnet). However, there is no response. Perhaps tellingly, 192.168.1.16 can reach 192.168.2.14 and it does respond. So for example I can ssh from 192.168.1.16 to 192.168.2.14, but not from 192.168.2.14 to 192.168.1.16.
I agree with Steve and my guess would be the gateway.
You can check by doing the packet capture again but on the WAN instead. You'll probably see the replies being sent out that way. -
I don't see reply packets on WAN. I did uncover more detail that is surprising, though:
Capturing on LAN, I see packets from 192.168.2.14 to 192.168.1.16 as expected. Capturing on the 192.168.1.16 machine, the same packets have the LAN mac address but a source IP of 192.1.68.1.2 -- that's the IP of a Netgear WiFi router running in AP mode (it doesn't run DHCP, and it isn't a gateway), which 192.168.1.16 is connected to. Of course, when 192.168.1.16 responds to 192.168.1.2 it gets ICMP host unreachable responses (with the correct Netgear mac address), and the TCP connection fails.
-
-
@steveits said in LAN access from VPN:
@ghazel but there is a rule on the VPN interface allowing access to lan?
Usually it’s:
Missing fw rule in arriving interfaceArriving interface (WG_VPN) has an 'Allow Any' rule.
Device fw doesn’t allow different subnet
Device fw is not enabled. (behavior is the same even with it enabled, since the source IP of the packets is 192.168.1.2 -- inside the subnet)
Device not using pfSense as gateway
Device is using pfSense 192.168.1.1 as a gateway.
@viragomann said in LAN access from VPN:
@ghazel
Seem a bit mysterious. Can't really believe.Maybe the Netgear is still doing source IP rewriting for sources that are outside the subnet?
I also suspect that the access is blocked by the destination device itself. This is the default behavior of an operating systems firewall.
Operating system firewall is off. Wireshark is seeing the packets and the responses.
-
@ghazel said in LAN access from VPN:
Maybe the Netgear is still doing source IP rewriting for sources that are outside the subnet?
It can do this only on traffic which passes through it.
Operating system firewall is off. Wireshark is seeing the packets and the responses.
Can you post this capture and tell us, on which device and NIC it was taken.
-
@viragomann said in LAN access from VPN:
@ghazel said in LAN access from VPN:
Maybe the Netgear is still doing source IP rewriting for sources that are outside the subnet?
It can do this only on traffic which passes through it.
The traffic is passing through it. Internet > WG_VPN > LAN > Netgear > 192.168.1.16
Operating system firewall is off. Wireshark is seeing the packets and the responses.
Can you post this capture and tell us, on which device and NIC it was taken.
pfSense_LAN.cap
192.168.1.16_en6.pcapngUploaded capture from pfSense on LAN interface 192.168.1.1 90:ec:77:32:cb:ad (which is plugged into the Netgear, 192.168.1.2 3c:37:86
fc:ae) and capture from target machine en6 192.168.1.16 8c:ae:4c:dd:4a:91 (plugged into that Netgear). -
@ghazel
Indeed, it's as you said before.Maybe the Netgear is still doing source IP rewriting for sources that are outside the subnet?
Obviously that's the case.
But it must no do this if it's running in AP mode as you mentioned. In AP mode all traffic through the device is on layer 2 only, no routing or natting / masqerading would be possible.
So there must be something wrong with the settings of the Netgear AP. You should re-check this.
-
@viragomann said in LAN access from VPN:
But it must no do this if it's running in AP mode as you mentioned. In AP mode all traffic through the device is on layer 2 only, no routing or natting / masqerading would be possible.
So there must be something wrong with the settings of the Netgear AP. You should re-check this.
Unfortunately in AP mode there are zero additional settings.
I could get a standalone switch for the wired devices to connect to LAN, but the wireless devices would still have the same problem.
Would it be practical to force WG_VPN to use a section of the 192.168.1.x range, so that wireguard peers appeared to be on the same subnet as the Netgear?
-
@ghazel said in LAN access from VPN:
Unfortunately in AP mode there are zero additional settings.
But there must be something wrong with it. Did you reboot or better reset the device?
Replacing the source IP with its own only makes if the wifi device is in router mode and redirects respond packets back to the origin source.
Can it operate as a router?
This could be a workaround.Would it be practical to force WG_VPN to use a section of the 192.168.1.x range, so that wireguard peers appeared to be on the same subnet as the Netgear?
You could masquerade the outgoing traffic on pfSense LAN interface and get the same. But if the switch replaces the source IP anyway, what I think, you win nothing.
-
@viragomann said in LAN access from VPN:
But there must be something wrong with it. Did you reboot or better reset the device?
Rebooted many times (thanks PG&E!). Have not tried a full reset, but will.
Can it operate as a router?
This could be a workaround.Yes, with the downside of double-NAT. Services that currently work fine with upnp/nat-pmp would stop working.
You could masquerade the outgoing traffic on pfSense LAN interface and get the same. But if the switch replaces the source IP anyway, what I think, you win nothing.
It seems like the switch does not replace the source IP if it's in the same subnet range (192.168.1.x). So if the wireguard peer got 192.168.1.114 or something like that (instead of 192.168.2.14), maybe that would work?
-
@ghazel said in LAN access from VPN:
It seems like the switch does not replace the source IP if it's in the same subnet range (192.168.1.x).
Yes, I had even the same idea. So maybe you could try to masquerade the traffic on pfSense.
You can do this with an outbound NAT rule. On the outbound NAT tab enable the hybrid mode first.
Then add a rule with this values:
interface: LAN
source: 192.168.2.0/24
destination: any
translation: interface addressThis replaces the source address with pfSense LAN address, so it's inside the subnet. Maybe this works.