CODEWITHSUNDEEP
checksumchef-infrarecipe
jamesc
jamesc

Reputation: 6328

Undefined method `checksum' for Chef::Resource::Directory

While running a recipe for Pyramid, the following is failing in chef-solo:

================================================================================
Error executing action `create` on resource 'directory[/etc/service/pyramid/supervise/status]'
================================================================================

NoMethodError
-------------
undefined method `checksum' for Chef::Resource::Directory

Resource Declaration:
---------------------
# In /var/chef-solo/cookbooks/project/recipes/pyramid.rb

 22:     directory "#{node['runit']['service_dir']}/pyramid#{dir}/supervise/#{f}" do
 23:       owner login
 24:       group login
 25:     end
 26:   end

Compiled Resource:
------------------
# Declared in /var/chef-solo/cookbooks/project/recipes/pyramid.rb:22:in `block (2 levels) in from_file'

directory("/etc/service/pyramid/supervise/status") do
  provider Chef::Provider::Directory
  action :create
  retries 0
  retry_delay 2
  path "/etc/service/pyramid/supervise/status"
  cookbook_name :project
  recipe_name "pyramid"
  owner "myusername"
  group "myusername"
  mode 420
end

Why is this getting "undefined method checksum" and what can I do about it?

This error doesn't appear to be consistent. Sometimes the chef run appears to hang. Other times, after multiple runs, chef is able to pass this point.

I'm running on a vanilla Ubuntu server 12.04LTS with chef installed with a deploy script:

sudo apt-get install -y ruby1.9.1 ruby1.9.1-dev make &&
sudo gem1.9.1 install chef ohai --no-rdoc --no-ri

This is the full Pyramid recipe:

login = node["user"]["login"]
home = node["user"]["home"]
app_root = node["user"]["app_root"]

runit_service "pyramid" do #, :template_name => "site" do
  template_name "pyramid"
  owner login
  group login
  options({
    :login => login,
    :app_root => app_root,
    :home => home,
    :config => "#{app_root}/config/gunicorn.conf.py"
  })
end

["", "/log"].each do |dir|
  directory "#{node['runit']['service_dir']}/pyramid#{dir}/supervise" do
    mode "0755"
  end
  %w(ok control status).each do |f|
    directory "#{node['runit']['service_dir']}/pyramid#{dir}/supervise/#{f}" do
      owner login
      group login
    end
  end
end

Upvotes: 2

Views: 2116

Answers (1)

jamesc
jamesc

Reputation: 6328

This was my solution - found after some hacking around plus the tip from @Draco-ater...

Where the recipe referenced directory("/etc/service/pyramid/supervise/status"), the status 'directory' is a file that already exists which was causing issues.

In addition, the script also tried to update the control and ok pipes as directories - which was also failing. I solved this by using bash and chmod:

bash "give_perms_pyramid" do
    user "chef"
    cwd "#{node['runit']['service_dir']}/pyramid#{dir}/supervise"
    code <<-EOH
        sudo chown myuser:root control
        sudo chmod g+rw control
    EOH

Maybe that will help someone.

Upvotes: 3

Related Questions