DNS problem

[michael-h-williamson]

Hi,

I have a computer with Ubuntu 20.04 that is not resolving names. It is configured for Network Address Translation for a Local Area Network. The computers on the LAN can access the internet through it and DNS works for those computers. On the NAT computer, I put in DNS IPs that should work using the GUI configuration. In fact, supplying a DNS IP to the dig command in a terminal window does work. For example:

# dig @8.8.8.8 www.yahoo.com

works. However, without providing the DNS IP to query, like so

# dig www.yahoo.com

it responds with "SERVFAIL"

; <<>> DiG 9.16.1-Ubuntu <<>> www.yahoo.com
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: SERVFAIL, id: 32672

;; Query time: 0 msec
;; SERVER: 127.0.0.53#53(127.0.0.53)

I tried switching on and off the "Automatic" option in the GUI for DNS with no luck. I can query systemd for the current DNS server and replies with the IP addresses I supplied (which all work):

Link 3 (enp0s31f6)
Current Scopes: DNS
DefaultRoute setting: yes
LLMNR setting: yes
MulticastDNS setting: no
DNSOverTLS setting: no
DNSSEC setting: no
DNSSEC supported: no
Current DNS Server: 192.168.0.1
DNS Servers: 192.168.0.1
8.8.8.8
8.8.4.4
fc7d:ed91::200::1
DNS Domain: ~.

When I don't supply an IP, it (dig) uses 127.0.0.53, which fails (in 0 msec).

The internet access for this computer is Hughesnet satellite. To configure it for NAT I enabled port forwarding and installed the package iptables-persistent and added an iptables rule based on instructions I found on the internet. DNS for this computer was working until I configured it for NAT. Also, in the Ubuntu network configuration GUI, all of the IPv6 stuff is in the default "Automatic" setting.

Does anyone know what needs to be changed to fix this?

Thanks,
-Mike

[SeijiSensei]

I can query systemd for the current DNS server and replies with the IP addresses I supplied (which all work):

Are those addresses entered in /etc/systemd/resolved.conf? If you enter them there, then run "sudo systemctl restart systemd-resolved", you might have better luck.

If this is a virtual machine, I see no reason to use iptables at all, especially if you're using NAT. The VM is not exposed directly to any system outside of the host computer.

If you need other computers on the local network to be able to see this virtual machine, you need to use "bridged networking" not NAT. In a bridged setup, the VM will get an address directly on the public network and be a peer to all the other machines.

[michael-h-williamson]

Thanks, I tried putting the line

DNS=192.168.0.1

in the file "/etc/systemd/resolved.conf" and rebooted, but by default it still uses 127.0.0.53. The computer is not a VM.

The file "/etc/resolv.conf" is a link pointing to "/run/systemd/resolve/stub-resolv.conf which contains the lines:

nameserver 127.0.0.53
options edns0 trust-ad

That is where it's getting the 127.0.0.53 address. The file says not to edit it directly, but to use the man page instructions for systemd-resolved.service, so that's what I'll try next.

[michael-h-williamson]

To answer my own question, I replaced the symlink /etc/resolv.conf with a file containing the DHCP supplied DNS address instead of 127.0.0.53, and that seems to work. I now interpret the instructions differently, after reading the systemd-resolved.service man page. The "Do not edit this file" comment refers to the /run/systemd/resolve/stub-resolv.conf file, not the symlink, which I think it is safe to replace with a regular file.