Vagrant: Error on installing Docker using the Docker provisioner

[thasmo]

The following happens when Vagrant creates the Docker host on Windows:

Bringing machine 'default' up with 'docker' provider...
==> default: Docker host is required. One will be created if necessary...
    default: Vagrant will now create or start a local VM to act as the Docker
    default: host. You'll see the output of the `vagrant up` for this VM below.
    default:
    default: Importing base box 'phusion/ubuntu-14.04-amd64'...
    default: Matching MAC address for NAT networking...
    default: Checking if box 'phusion/ubuntu-14.04-amd64' is up to date...
    default: Setting the name of the VM: Docker
    default: Clearing any previously set forwarded ports...
    default: Clearing any previously set network interfaces...
    default: Preparing network interfaces based on configuration...
    default: Adapter 1: nat
    default: Forwarding ports...
    default: 22 => 2222 (adapter 1)
    default: Running 'pre-boot' VM customizations...
    default: Booting VM...
    default: Waiting for machine to boot. This may take a few minutes...
    default: SSH address: 127.0.0.1:2222
    default: SSH username: vagrant
    default: SSH auth method: private key
    default: Warning: Connection timeout. Retrying...
    default: Machine booted and ready!
    default: Checking for guest additions in VM...
    default: Mounting shared folders...
    default: /vagrant => S:/Project/vagrant.docker-host
    default: Running provisioner: docker...
    default: Installing Docker (latest) onto machine...
    default: Configuring Docker to autostart containers...
==> default: Syncing folders to the host VM...
    default: Mounting shared folders...
    default: /var/lib/docker/docker_1402252891_7417 => S:/Project/docker.test
A Docker command executed by Vagrant didn't complete successfully!
The command run along with the output from the command is shown
below.

Command: "docker" "ps" "-a" "-q" "--no-trunc"

Stderr: 2014/06/08 18:41:49 Get http:///var/run/docker.sock/v1.12/containers/json?all=1: dial unix /var/run/docker.sock: permission denied

Stdout:

This is the Vagrantfile for the Docker host:

Vagrant.configure(2) do |config|
  config.vm.box = 'phusion/ubuntu-14.04-amd64'
  config.vm.provision 'docker'

  config.vm.provider 'virtualbox' do |provider|
    provider.gui = false
    provider.name = 'Docker'
    provider.customize ['modifyvm', :id, '--cpus', 2]
    provider.customize ['modifyvm', :id, '--acpi', 'on']
    provider.customize ['modifyvm', :id, '--memory', 2048]
    provider.customize ['modifyvm', :id, '--cpuexecutioncap', '100']
    provider.customize ['modifyvm', :id, '--natdnshostresolver1', 'on']
    provider.customize ['modifyvm', :id, '--natdnsproxy1', 'on']
  end

  ['vmware_fusion', 'vmware_workstation'].each do |vmware|
    config.vm.provider vmware do |provider|
      provider.gui = false
      provider.vmx['displayName'] = 'Docker'
      provider.vmx['numvcpus'] = 2
      provider.vmx['memsize'] = 2048
    end
  end

  config.ssh.forward_agent = true
end

Nevertheless the Docker host seems to be running afterwards. Haven't tested further yet.

[thasmo]

Could be a Docker problem: https://github.com/dotcloud/docker/issues/5314

[thasmo]

The vagrant user seems to be member of the docker group, so this shouldn't be the problem. Weird.

[mitchellh]

Ah, the issue is that you need to log out and log back in. The Docker provisioner i nstaller doesn't do this because all the solutions are pretty hacky. But google how to do that in a shell and it'll work.

The root cause: group membership changes only take effect on log out and log in.

[mitchellh]

We might be able to automate this in the future in some hacky way, actually. Let me reopen and tag as enhancement.

[dklotz]

I'm using a custom VM referenced with vagrant_vagrantfile in my docker-based project, that proxy vm installs docker (and adds the vagrant user to the group docker) with a normal shell provisioning script defined in its Vagrantfile. Yet when I 'vagrant up' my real (i.e. not the proxy-vm) docker-vagrant-project for the first time, the proxy vm is started yet creating the docker container fails with the following:

A Docker command executed by Vagrant didn't complete successfully!
The command run along with the output from the command is shown below.
Command: "docker" "ps" "-a" "-q" "--no-trunc"
Stderr: 2014/08/08 16:46:19 Get http:///var/run/docker.sock/v1.13/containers/json?all=1: dial unix /var/run/docker.sock: permission denied

The proxy VM keeps running after that and when I try to start the docker-based VM again, it works fine. Where should I add the "logging in / out" workaround mentioned in the comments here? As a provisioning script in the Vagrantfile of my proxy-base-docker-VM? Or somehow in the Vagrantfile of the "real" dockerized project?

[dklotz]

@mitchellh could you maybe document the workaround for the time being somewhere? I know how to login/logout/newgrp in a script, I just haven't been able to figure out where to put this in the vagrant/docker context. (also: didn't you want to reopen this issue?)

[dklotz]

@mitchellh this issue is still existent and makes using vagrant with a custom docker host VM really annoying. Could you at least document the login/logout workaround thats supposed to work somewhere? I haven't been able to configure vagrant with a custom base VM that doesn't error out on the first try with this error. Trying again helps, but having to tell all our developers "just execute the command a second time" isn't really a good workaround...

[mitchellh]

@dklotz I think we can probably do better and detect these errors and maybe show a human-friendly error. Tagged as docs and reopening.

[kmmbvnr]

Isn't this the solution that works? Put it right after config.vm.provision 'docker' line

  # The following line terminates all ssh connections. Therefore
  # Vagrant will be forced to reconnect.
  # That's a workaround to have the docker command in the PATH
  config.vm.provision "shell", inline:
     "ps aux | grep 'sshd:' | awk '{print $2}' | xargs kill"

[dklotz]

@kmmbvnr thanks, yes, that seems to help! although I'm unable to really verify it thanks to another docker-related issue (#4723)

[sylus]

I can say that the fix provided by @kmmbvnr solved this issue for me :)

[cjerdonek]

My Vagrant VM is in a similar state:

No matter what command I run: vagrant status, vagrant destroy, vagrant ssh, etc, I always get the "A Docker command executed by Vagrant didn't complete successfully!" message. Is there any way to get this "unstuck" without completely wiping the .vagrant directory and starting over?

[sethvargo]

Hey everyone,

The next version of Vagrant will allow you to use your own custom Docker host. As Mitchell said, we could build a really hacky solution to somehow force a user-login-logout, but none of the solutions are very good or cross-compatible. The solution @kmmbvnr posted above is a really great option. Thanks! 😄