fgrehm/vagrant-lxc

`stack level too deep` when trying to use 'ansible_local' to provision

Closed this issue · 5 comments

krtek4 commented

Hello,

I am having a strange issue. I wanted to try the new ansible_local provisioner in my LXC boxes and I ran into the following issue :

± vagrant up  
Bringing machine 'default' up with 'lxc' provider...
==> default: Importing base box 'jessie64-lxc'...
==> default: Setting up mount entries for shared folders...
    default: /vagrant => /home/krtek/Repositories/liip/rawbot-virtualization/playground
==> default: Starting container...
==> default: Waiting for machine to boot. This may take a few minutes...
/opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/loggerfactory.rb:76:in `module_eval': stack level too deep (SystemStackError)
    from /opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/loggerfactory.rb:76:in `set_log'
    from /opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/loggerfactory.rb:39:in `block in toggle_methods'
    from /opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/loggerfactory.rb:35:in `each'
    from /opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/loggerfactory.rb:35:in `toggle_methods'
    from /opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/loggerfactory.rb:19:in `define_methods'
    from /opt/vagrant/embedded/gems/gems/log4r-1.1.10/lib/log4r/logger.rb:37:in `initialize'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/lib/vagrant/action/warden.rb:21:in `new'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/lib/vagrant/action/warden.rb:21:in `initialize'
     ... 1554 levels...
    from /opt/vagrant/embedded/lib/ruby/2.2.0/timeout.rb:32:in `catch'
    from /opt/vagrant/embedded/lib/ruby/2.2.0/timeout.rb:103:in `timeout'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/communicators/ssh/communicator.rb:42:in `wait_for_ready'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/lib/vagrant/action/builtin/wait_for_communicator.rb:16:in `block in call'

 ± vagrant destroy
/opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/guest.rb:40:in `initialize_dup': stack level too deep (SystemStackError)
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/guest.rb:40:in `initialize'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/guest.rb:40:in `new'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/guest.rb:40:in `check_path'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/guest.rb:53:in `check_path_is_a_file'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/base.rb:78:in `validate'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/provisioners/ansible/config/guest.rb:32:in `validate'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/kernel_v2/config/vm.rb:730:in `block in validate'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/kernel_v2/config/vm.rb:720:in `each'
     ... 8275 levels...
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/plugins/commands/destroy/command.rb:30:in `execute'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/lib/vagrant/cli.rb:42:in `execute'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/lib/vagrant/environment.rb:302:in `cli'
    from /opt/vagrant/embedded/gems/gems/vagrant-1.8.1/bin/vagrant:174:in `<main>'

What I don't really understand is that the exception is raised in two different places when doing an up or a destroy. AFAIK, provision generate the same error as destroy. The file and lines are consistent across multiple runs.

FYI, here's a minimal Vagrantfile that triggers the error for me :

# -*- mode: ruby -*-
# vi: set ft=ruby :

VAGRANTFILE_API_VERSION = "2"

Vagrant.require_version ">= 1.8.1"

Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
    config.vm.box = "glenux/jessie64-lxc"

    config.vm.provision 'ansible_local' do |ansible|
        ansible.playbook = 'playbook.yml'
    end
end

I tested using the VirtualBox provider, and everything worked fine, so I am assuming the error is some sort of incompatibility between vagrant-lxc and the new provisioner.

Here are some more information about my setup :

 $ vagrant plugin list
vagrant-hostmanager (1.6.1)
vagrant-lxc (1.2.1)
vagrant-share (1.1.5, system)
$ vagrant -v
Vagrant 1.8.1
$ lxc-start --version
1.0.8

Full debug log of vagrant provision : http://pastebin.com/jUmz9zpw

krtek4 commented

Any input on this ? Can I help in any way ?

ccope commented

What happens if you try bringing up the box without the ansible block in your Vagrantfile?

krtek4 commented

It works fine... in fact it works great with :

  • no provisioner
  • the ansible provisioner
  • the shell provisioner

Probably others, the only one having the issue is the ansible_local provisioner.

I should also add that the ansible_local provisioner work great on VirtualBox, I only have the issue with the LXC provider, this is why I opened this here and not and Vagrant itself.

pradeepk commented

Same issue with aws provider (vagrant-aws plugin). See hashicorp/vagrant#6984

krtek4 commented

Seems this was fixed on the Vagrant side in the referenced issue.

I am then closing this :)