Sometimes, Docker‘s internet connectivity won’t be working properly, which can lead to a number of obscure errors with your applications. In my experience, this is usually because DNS lookups are failing in Docker images.

If you know it’s a DNS problem and you’re in a hurry, jump straight to the system-wide solution.

Is DNS the problem?

Fortunately it’s easy to test Docker’s DNS.

First, check that basic internet connectivity is working by pinging a public IP address. It should succeed, giving you output similar to this:

$ docker run busybox ping -c 1  # Ping a London-based NASA root nameserver
PING ( 56 data bytes
64 bytes from seq=0 ttl=53 time=113.866 ms

--- ping statistics ---
1 packets transmitted, 1 packets received, 0% packet loss
round-trip min/avg/max = 113.866/113.866/113.866 ms

But now try resolving the domain

$ docker run busybox nslookup
Address 1:
nslookup: can't resolve ''

If it fails as shown above then there is a problem resolving DNS.


By default, if Docker can’t find a DNS server locally defined in your /etc/resolv.conf file, containers will default to using Google’s public DNS server8.8.8.8, to resolve DNS.

In some networks, like Canonical’s London office network where I work, the administrators intentionally block the use of public DNS servers to encourage people to use the network’s own DNS server.

In this case, Docker containers using the default configuration won’t be able to resolve DNS, rendering the internet effectively unuseable from within those containers.

I’ve filed a bug about this issue, although I don’t yet know when or if it might be addressed.

The quick fix: Overriding Docker’s DNS

Fortunately, it’s fairly easy to directly run a docker container with a custom DNS server.

Discover the address of your DNS server

You can find out what network’s DNS server from within Ubuntu as follows:

$ nmcli dev show | grep 'IP4.DNS'

Run Docker with the new DNS server

To run a docker container with this DNS server, provide the --dns flag to the run command. For example, let’s run the command we used to check if DNS is working:

$ docker run --dns busybox nslookup
Address 1:
Address 1: 2a00:1450:4009:811::200e
Address 2:

And that’s what success looks like.

The permanent system-wide fix

The above solution is all very well if you’re only temporarily inside a restrictive network and you only need to run containers directly.

However, most of the time you’ll want this to work by default and keep working on your system, and for any other programs that rely on Docker.

Update the Docker daemon

To achieve this, you need to change the DNS settings of the Docker daemon. You can set the default options for the docker daemon by creating a daemon configuration file at /etc/docker/daemon.json.

You should create this file with the following contents to set two DNS, firstly your network’s DNS server, and secondly the Google DNS server to fall back to in case that server isn’t available:


    "dns": ["", ""]

Then restart the docker service:

sudo service docker restart

Testing the fix

Now you should be able to ping successfully from any Docker container without explicitly overriding the DNS server, e.g.:

$ docker run busybox nslookup
Address 1:
Address 1: 2a00:1450:4009:811::200e
Address 2:

Please Note: This post is collected from site. I am just posting the solution here as my future reference. All the credit and copyrights goes to the mentioned site.