This actually reminded me that there are some UI bits in 0.1.6 that are not available on 2.5.2/21.05... I should probably re-version 0.1.6 to be 0.2.0 to reflect that change. Hmmm

@luckman212 It's now appearing in the pfSense package manager, at least for 2.6.0-devel.

@bcruze I'm not sure what outbound NAT has to do with redirecting DNS queries. Can you explain it to me please? What should I change to redirect the VPN interface's DNS queries through the VPN gateway? Thank you

@dma_pf said in DNS Not Working With Phone As Peer:

@bingo600 @GenericStudent

Thanks to both of you for all of your help. I spent several hours this morning working on this issue and finally got it resolved. It was a combination of 3 different issues that resolved it.

Before I get to the solution I want to clarify that the setup I have is using a dedicated assigned interface with a gateway assigned to it for the remote access tunnel. Like this:

496c0552-b686-48e8-a473-ed95c64c0ead-image.png

181b3442-cdf2-4a90-b875-77cbdcbb8e3a-image.png

I found that first issue I had was the NAT Rule I posted above was not needed:
a5d6906c-2253-4acc-9819-437e134e3175-image.png

The second issue I had was the NAT Rule I posted above was also not needed:
d47715d8-e06f-4300-bef7-254f91def188-image.png
The reason it is not needed is because the 10.0.9.0 network is already know to pfsense through the assignment of that network to the wireguard interface and gateway. If there was no local pfsense interface assigned to that wireguard tunnel then the NAT rule would have been required.

The third issue was exactly what @bingo600 pointed out. I did need to create a rule to allow the 10.0.9.0 network to assess the DNS resolver like this:

62e2c01a-58db-42f3-b0a0-7414fceaa19d-image.png

I am very perplexed as to why that allow rule had to be created. The setting I posted above, d707a8d1-46f5-481e-8b49-caf358dfdbb1-image.png
should have allowed the DNS queries as the 10.0.9.0 network is a local pfsense network. The pfsense cleary indicates that by selecting "All" there should not need an allow rule. Per the pfsense documentation:

ab33f227-2fc2-4c90-a98e-9f56e85c129e-image.png

5ac51888-b59e-4150-8021-6bd37b34c152-image.png

I'm obviously misunderstanding something about why that access rule is required. If you can help me understand that better I'd greatly appreciate it.

Thank you guys for all of your help. I've been trying to figure this issue out for several weeks and your input got me pointed in the right direction to get it resolved. 😀

Thank you so much for this post. I was experiencing exactly the same issue and you helped to fix it!

@cmcdonald said in WireGuard Road Warrior setup:

This isn't necessary for assigned tunnel interfaces as pfSense already appends these subnets to the Unbound config, but for unassigned tunnel interfaces this additional "step" is required. It should be automatic.

@hulleyrob said in WireGuard Road Warrior setup:

yes it was

I had the exact same thing happen to me. I have a Wireguard tunnel installed as an interface for my road warrior set up. I could not get it to resolve until I created in ACL for Unbound. This drove me nuts for days, the details are all in this thread https://forum.netgate.com/topic/165818/dns-not-working-with-phone-as-peer?_=1637700225107

It seems to me that there is an issue/bug where the Wireguard tunnel is not recognized by Unbound and therefore Unbound does not see the interface as an internal network and therefore requires the ACL.

@cmcdonald Thanks for the info.

I took the plunge and updated this "lab" 2100 to 22.01-devel from 21.09-RC and, sure enough, WG was also updated to 0.1.5_3.

I feel you because I am dealing with a similar situation with pf wireguard. PIA is my VPN provider. Their linux app on ubuntu VM runs fine. PIA also has a tool to generate wireguard conf file to work with wg-quick on ubuntu. No problem.

I generate the details in my ubuntu wireguard conf and enter the info in pf gui. Mapping is private key for tunnel. Endpoint and public key for peer. Address for opt interface and routing.

The pf wireguard peer does not always handshake. The first time pf wireguard connected to PIA it was perfect. The connection dropped after 2 weeks. Now the connection has long ping times and very slow.

Is there a way to automate mapping linux wireguard conf to pf wireguard tunnel and peer conf?

@kodols Sure.

You would create a 'site-to-site' style tunnel between pfSense and Mullvad, and then a second tunnel using the road warrior model. Then it just becomes a matter of setting up policy routing and firewall rules to accomplish the desired outcome.

@xiki It isn't clear if and how this is related to pfSense.

@cmcdonald said in WireGuard Widget:

@ciscox Noted! Thanks. That is particular useful for widgets that can be added multiple times to the dashboard, so it might also be worth allow multiple widgets with each filterable by specific tunnels.

Hi,
Exactly what I was thinking.:) 👍 😊

@bassopt Take another peak now, updated package should be available.

https://files01.netgate.com/pfSense_v2_5_2_amd64-pfSense_v2_5_2/All/pfSense-pkg-WireGuard-0.1.5_3.txz

@dma_pf Good call! I'll try it. Thank you so much for your help.

@securvark I had the same issue. This worked for me as well. Thanks for posting.

@mooncaptain You only need the rules on one interface for each FW. Sounds like you are good to go.

Well after these whole hours I might say that the issue can be on the HAProxy side, but I cannot say why ...
HAproxy is configured to the WAN IP on port 443/80.
If I disable the HAproxy I can connect to the web servers, but not if HAproxy is enabled.

Why?
Shouldn't the 'routing' between the LAN and VPN interfaces be dealt with without passing thru the WAN?
It seems that traffic goes to WAN ... and is being intercepted by the HAproxy .
I'm confused .. or I'm missing a big issue in here ...

JG