27

On Ubuntu 14.04 I'm using dnsmasq to resolve wildcard example.com domains to the local machine (desktop computer at home).

After much reading around I cannot for the life of me sort out the same on a clean install of 18.04.

All I want to achieve at the moment is:

  • for ping example.com to ping 127.0.0.1 and not 93.184.216.34;
  • for ping anysubdomain.example.com to also ping 127.0.0.1;
  • and for ping google.com to ping the real google.com via the router/IP DNS.

Surely this must be simple, even trivial?

But I'm stumped. I can get the example.com's resolving but only at the expense of breaking everything else.

How can I do this?

Improve this question
2
  • 1
    Why not just install dnsmasq like you did on 14.04?
    – vidarlo
    May 4, 2018 at 4:58
  • 1
    @vidarlo Well of course that was the first thing I did, as I indicated in my question. But unlike in 14.04 that results in a "failed to create listening socket for port 53: Address already in use" error. The answer I accepted works fine without installing dnsmasq explicitly.
    – Nick Rice
    May 5, 2018 at 0:34

5 Answers 5

Reset to default
57
+50

Here are the steps for ubuntu since 18.04. It's a little bit long since systemd-resolved does not play very well with NetworkManager when configured with dnsmasq.

Yet I still recommend starting dnsmasq from NetworkManager, because network connectivity changes (WIFI, wired, ...) will be handled transparently.

Enable dnsmasq in NetworkManager

Edit the file /etc/NetworkManager/NetworkManager.conf, and add the line dns=dnsmasq to the [main] section, it will look like this :

[main]
plugins=ifupdown,keyfile
dns=dnsmasq

[ifupdown]
managed=false

[device]
wifi.scan-rand-mac-address=no

Let NetworkManager manage /etc/resolv.conf

sudo rm /etc/resolv.conf ; sudo ln -s /var/run/NetworkManager/resolv.conf /etc/resolv.conf

Configure example.com

echo 'address=/.example.com/127.0.0.1' | sudo tee /etc/NetworkManager/dnsmasq.d/example.com-wildcard.conf

Reload NetworkManager and testing

NetworkManager should be reloaded for the changes to take effect.

sudo systemctl reload NetworkManager

Then we can verify that we can reach some usual site :

dig askubuntu.com +short
151.101.129.69
151.101.65.69
151.101.1.69
151.101.193.69

And lastly verify that the example.com and subdomains are resolved as 127.0.0.1:

dig example.com askubuntu.example.com a.b.c.d.example.com +short
127.0.0.1
127.0.0.1
127.0.0.1
Improve this answer
12
  • 1
    When NetworkManager uses dns=dnsmasq, it should tell systemd-resolved to use dnsmasq, could be automatic.
    – pim
    May 4, 2018 at 11:50
  • 1
    Thank you, this worked perfectly! It seems I can't give you the bounty for another hour, and as it's gone one in the morning here I'm off to bed now and will do that when I switch back on tomorrow.
    – Nick Rice
    May 5, 2018 at 0:14
  • 2
    Awesome. I've read many different techniques. This is the one that finally worked.
    – Redsandro
    Jul 26, 2018 at 9:56
  • 3
    For those who prefer to revert back to systemd-resolved, /etc/resolv.conf points to /run/systemd/resolve/stub-resolv.conf by default.
    – Pothi Kalimuthu
    Aug 8, 2018 at 4:30
  • 1
    Works on Ubuntu 20.04 as well
    – sola
    Aug 5, 2020 at 20:01
2

First make sure in /etc/NetworkManager/NetworkManager.conf the following line is not present or commented out:

dns=dnsmasq

Restart NetworkManager:

sudo systemctl restart NetworkManager

Make sure the NetworkManager-controlled dnsmasq is not running anymore, either by killing the process or rebooting your system.

Then install dnsmasq:

sudo apt install dnsmasq

Add the following to /etc/dnsmasq.d/example.com:

address=/example.com/127.0.0.1

Restart dnsmasq:

sudo systemctl restart dnsmasq

Now you should have a wildcard dns override for example.com.

Improve this answer
8
  • 1
    I couldn't get this to work. Immediately following a clean minimal install (with the "erase disk" option) the sudo apt install dnsmasq gives a "failed to create listening socket for port 53: Address already in use" error. I continued to the end anyway in case a later step resolved that, but it didn't and this didn't work.
    – Nick Rice
    May 5, 2018 at 0:11
  • Then the dnsmasq that is started by NetworkManager was still running. So it should first be deactivated, eventually killed, and only then the dnsmasq package should be installed.
    – Sebastian Stark
    May 5, 2018 at 0:14
  • Thanks, Sebastian. However I already accepted pim's answer which worked flawlessly, so I'm able to move on now without playing around with it any more.
    – Nick Rice
    May 5, 2018 at 0:25
  • @NickRice No problem with this, but maybe others want to try
    – Sebastian Stark
    May 5, 2018 at 1:19
  • Yes of course, Sebastian. I did try your answer first because it was the simplest, and if all it needs is an added initial step then it's a very good one.
    – Nick Rice
    May 5, 2018 at 7:41
1

Based on pim's answer, I have created a gist containing a simple Bash script which uses DNSMasq in combination with NetworkManager to setup the .test domain pointing to 127.0.0.1

See the gist at https://gist.github.com/archan937/d35deef3b1f2b5522dd4b8f397038d27.

You can execute the script with the following command:

curl -sL https://gist.githubusercontent.com/archan937/d35deef3b1f2b5522dd4b8f397038d27/raw/setup-dnsmasq.sh | sudo bash

Cheers!

Improve this answer
3
  • This is awesome, but it doesn't work for me. It just ends up with could not ping foo.test
    – SimaPro
    Jan 20, 2021 at 20:41
  • Hey, @SimaPro. I increased the ping timeout to 1 second. Does that solve your issue?
    – Paul Engel
    Jan 21, 2021 at 21:56
  • Successful after I forked + updated the service restart syntax (for Ubuntu 22.04)
    – LittleBigDev
    Dec 28, 2022 at 9:20
0

This won't be as simple as editing the hosts file. You have a couple of options:

This python DNS proxy that will handle wildcards in /etc/hosts

Using DNSmasq

Improve this answer
2
  • Thanks. Your link to the DNSmasq question is what I am doing in 14.04. Just doing the same in a fresh 18.04 installation isn't working due to a port conflict. So out of the box something else needs doing compared to that. Looking at all that code needed for the python DNS proxy, I can't believe all that is necessary. It wasn't before except, I suppose, for those who insisted on using /etc/hosts.
    – Nick Rice
    May 4, 2018 at 21:20
  • Can you do netstat -tulpn to check what's using the port? EDIT: Nevermind, saw the accepted answer. the problem is systemd-resolve.
    – Harikrishnan R
    May 10, 2018 at 13:16
0

It is possible to use dnsmasq-base that is already pre-installed in Ubuntu instead of dnsmasq for the same purpose.

So don't install dnsmasq but do all steps from here https://askubuntu.com/a/1031896/211368

If you install dnsmasq it will collide with already running dnsmasq-base and you'll get an error:
failed to create listening socket for port 53: Address already in use

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .