Docker build “Could not resolve 'archive.ubunt

问题:

I've been trying to run Docker build on various files which previously worked before, which are now no longer working.

As soon as the Docker file included any line that was to install software it would fail with a message saying that the package was not found.

RUN apt-get -y install supervisor nodejs npm

The common message which showed up in the logs was

Could not resolve 'archive.ubuntu.com'

Any idea why any software will not install?

回答1:

Uncommenting DOCKER_OPTS="--dns 8.8.8.8 --dns 8.8.4.4" in /etc/default/docker as Matt Carrier suggested did NOT work for me. Nor did putting my corporation's DNS servers in that file. But, there's another way (read on).

First, let's verify the problem:

$ docker run --rm busybox nslookup google.com   # takes a long time
nslookup: can't resolve 'google.com'   # <--- appears after a long time
Server:    8.8.8.8
Address 1: 8.8.8.8

If the command appears to hang, but eventually spits out the error "can't resolve 'google.com'", then you have the same problem as me.

The nslookup command queries the DNS server 8.8.8.8 in order to turn the text address of 'google.com' into an IP address. Ironically, 8.8.8.8 is Google's public DNS server. If nslookup fails, public DNS servers like 8.8.8.8 might be blocked by your company (which I assume is for security reasons).

You'd think that adding your company's DNS servers to DOCKER_OPTS in /etc/default/docker should do the trick, but for whatever reason, it didn't work for me. I describe what worked for me below.

SOLUTION:

On the host (I'm using Ubuntu 16.04), find out the primary and secondary DNS server addresses:

$ nmcli dev show | grep 'IP4.DNS'
IP4.DNS[1]:              10.0.0.2
IP4.DNS[2]:              10.0.0.3

Using these addresses, create a file /etc/docker/daemon.json:

$ sudo su root
# cd /etc/docker
# touch daemon.json

Put this in /etc/docker/daemon.json:

{                                                                          
    "dns": ["10.0.0.2", "10.0.0.3"]                                                                           
}     

Exit from root:

# exit

Now restart docker:

$ sudo service docker restart

VERIFICATION:

Now check that adding the /etc/docker/daemon.json file allows you to resolve 'google.com' into an IP address:

$ docker run --rm busybox nslookup google.com
Server:    10.0.0.2
Address 1: 10.0.0.2
Name:      google.com
Address 1: 2a00:1450:4009:811::200e lhr26s02-in-x200e.1e100.net
Address 2: 216.58.198.174 lhr25s10-in-f14.1e100.net

REFERENCES:

I based my solution on an article by Robin Winslow, who deserves all of the credit for the solution. Thanks, Robin!

"Fix Docker's networking DNS config." Robin Winslow. Retrieved 2016-11-09. https://robinwinslow.uk/2016/06/23/fix-docker-networking-dns/

回答2:

After much headache I found the answer. Could not resolve 'archive.ubuntu.com' can be fixed by making the following changes:

1. Uncomment the following line in /etc/default/docker
DOCKER_OPTS="--dns 8.8.8.8 --dns 8.8.4.4"

2. Restart the Docker service sudo service docker restart

3. Delete any images which have cached the invalid DNS settings.

4. Build again and the problem should be solved.

Credit goes to Andrew SB

回答3:

I believe that Matt Carrier's answer is the correct solution for this problem. However, after implementing it, I still observed the same behavior: could not resolve 'archive.ubuntu.com'.

This led me to eventually find that the network I was connected to was blocking public DNS. The solution to this problem was to configure my Docker container to use the same name server that my host (the machine from which I was running Docker) was using.

How I triaged:

1. Since I was working through the Docker documentation, I already had an example image installed on my machine. I was able to start a new container to run that image and create a new bash session in that container: docker run -it docker/whalesay bash
2. Does the container have an Internet connection?: ping 172.217.4.238 (google.com)
3. Can the container resolve hostnames? ping google.com

In my case, the first ping resulted in responses, the second did not.

How I fixed:

Once I discovered that DNS was not working inside the container, I verified that I could duplicate the same behavior on the host. nslookup google.com resolved just fine on the host. But, nslookup google.com 8.8.8.8 or nsloookup google.com 8.8.4.4 timed out.

Next, I found the name server(s) that my host was using by running nm-tool (on Ubuntu 14.04). In the vein of fast feedback, I started up the example image again, and added the IP address of the name server to the container's resolv.conf file: sudo vi /etc/resolv.conf. Once saved, I attempted the ping again (ping google.com) and this time it worked!

Please note that the changes made to the container's resolv.conf are not persistent and will be lost across container restarts. In my case, the more appropriate solution was to add the IP address of my network's name server to the host's /etc/default/docker file.

回答4:

I run into the same problem, but neiter uncommenting /etc/default/docker dns entries nor editing the /etc/resolv.conf in the build container or the /etc/docker/daemon.json helps for me.

But after I build with the option --network=host the resolving was fine again.

docker build --network=host -t my-own-ubuntu-like-image .

Maybe this will help someone again.

回答5:

After adding local dns ip to default docker file it started working for me... please find the below steps...

$ nm-tool # (will give you the dns IP)

DNS : 172.168.7.2

$ vim /etc/default/docker # (uncomment the DOCKER_OPTS and add DNS IP)
DOCKER_OPTS="--dns 172.168.7.2 --dns 8.8.8.8 --dns 8.8.4.4"

$ rm `docker ps --no-trunc -aq` # (remove all the containers to avoid DNS cache)

$ docker rmi $(docker images -q) # (remove all the images)

$ service docker restart #(restart the docker to pick up dns setting)

Now go ahead and build the docker... :)

回答6:

For anyone who is also having this problem, I solved my problem by editing the /etc/default/docker file, as suggested by other answers and questions. However I had no idea what IP to use as the DNS.

It was only after a while I figured out I had to run ifconfig docker on the host to show the IP for the docker network interface.

docker0   Link encap:Ethernet  Endereço de HW 02:42:69:ba:b4:07  
          inet end.: 172.17.0.1  Bcast:0.0.0.0  Masc:255.255.0.0
          endereço inet6: fe80::42:69ff:feba:b407/64 Escopo:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Métrica:1
          pacotes RX:8433 erros:0 descartados:0 excesso:0 quadro:0
          Pacotes TX:9876 erros:0 descartados:0 excesso:0 portadora:0
          colisões:0 txqueuelen:0 
          RX bytes:484195 (484.1 KB) TX bytes:24564528 (24.5 MB)

It was 172.17.0.1 in my case. Hope this helps anyone who is also having this issue.

回答7:

I just wanted to add a late response for anyone coming across this issue from search engines.

Do NOT do this: I used to have an option in /etc/default/docker to set iptables=false. This was because ufw didn't work (everything was opened even though only 3 ports were allowed) so I blindly followed the answer to this question: Uncomplicated Firewall (UFW) is not blocking anything when using Docker and this, which was linked in the comments

I have a very low understanding of iptables rules / nat / routing in general, hence why I might have done something irrational.

Turns out that I've probably misconfigured it and killed DNS resolution inside my containers. When I ran an interactive container terminal: docker run -i -t ubuntu:14.04 /bin/bash

I had these results:

root@6b0d832700db:/# ping google.com
ping: unknown host google.com

root@6b0d832700db:/# cat /etc/resolv.conf
search online.net
nameserver 8.8.8.8
nameserver 8.8.4.4

root@6b0d832700db:/# ping 8.8.8.8
PING 8.8.8.8 (8.8.8.8) 56(84) bytes of data.
64 bytes from 8.8.8.8: icmp_seq=1 ttl=56 time=1.76 ms
64 bytes from 8.8.8.8: icmp_seq=2 ttl=56 time=1.72 ms

Reverting all of my ufw configuration (before.rules), disabling ufw and removing iptables=false from /etc/default/docker restored the DNS resolution functionality of the containers.

I'm now looking forward to re-enable ufw functionality by following these instructions instead.

回答8:

I found this answer after some Googleing. I'm using Windows, so some of the above answers did not apply to my file system.

Basically run:

docker-machine ssh default
echo "nameserver 8.8.8.8" > /etc/resolv.conf

Which just overwrites the existing nameserver used with 8.8.8.8 I believe. It worked for me!

回答9:

I have the same issue, and tried the steps mentioned, but seems none works until refresh the network settings.

The steps:

1. As mentioned, add DOCKER_OPTS="--dns 8.8.8.8 --dns 8.8.4.4 --ip-masq=true" to /etc/default/docker.
2. Manually flush the PREROUTING table contents using the iptables -t nat -F POSTROUTING . After running this, restart docker and it will initialize the nat table with the new IP range.

回答10:

Same issue for me (on Ubuntu Xenial).

• docker run --dns ... for containers worked.
• Updating docker daemon options for docker build (docker-compose etc.) did not work.

After analyzing the docker logs (journalctl -u docker.service) if found some warning about bad resolvconf applied.

Following that i found that our corporate nameservers were added to the network interfaces but not in resolvconf.

Applied this solution How do I configure my static DNS in interfaces? (askubuntu), i.e. adding nameservers to /etc/resolvconf/resolv.conf.d/tail

After updating resolvconf (or reboot).

bash docker run --rm busybox nslookup google.com

worked instantly.

All my docker-compose builds are working now.

回答11:

I got same issue today, I just added line below to /etc/default/docker

DOCKER_OPTS="--dns 172.18.20.13 --dns 172.20.100.29 --dns 8.8.8.8"

and then I restarted my Laptop.

In my case restarting docker daemon is not enough for me, I have to restart my Laptop to make it work.