Merge pull request #585 from Capgemini/coreos

Initial commit of work in progress on CoreOS

Vagrantfile

@@ -1,44 +1,66 @@
 # -*- mode: ruby -*-
 # vi: set ft=ruby :
 require 'yaml'
+require 'fileutils'
 
 base_dir = File.expand_path(File.dirname(__FILE__))
 conf = YAML.load_file(File.join(base_dir, "vagrant.yml"))
 groups = YAML.load_file(File.join(base_dir, "ansible-groups.yml"))
 
-require File.join(base_dir, "vagrant_helper")
+CONFIG_HELPER = File.join(base_dir, "vagrant_helper.rb")
+CLOUD_CONFIG_PATH = File.join(base_dir, "user-data")
+
+if File.exist?(CONFIG_HELPER)
+  require CONFIG_HELPER
+end
 
 # Vagrantfile API/syntax version. Don't touch unless you know what you're doing!
 VAGRANTFILE_API_VERSION = "2"
-Vagrant.require_version ">= 1.7.0"
+Vagrant.require_version ">= 1.8.0"
 
 Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
-  # if you want to use vagrant-cachier,
-  # please install vagrant-cachier plugin.
-  if Vagrant.has_plugin?("vagrant-cachier")
-    config.cache.enable :apt
-    config.cache.scope = :box
+
+  config.vm.box              = "coreos-%s" % conf['coreos_update_channel']
+  #config.vm.box_version      = ">= 877.1.0"
+  config.vm.box_url          = "http://%s.release.core-os.net/amd64-usr/current/coreos_production_vagrant.json" % conf['coreos_update_channel']
+
+  if Vagrant.has_plugin?('vagrant-vbguest') then
+    config.vbguest.auto_update = false
   end
 
-  # throw error if vagrant-hostmanager not installed
-  unless Vagrant.has_plugin?("vagrant-hostmanager")
-    raise "vagrant-hostmanager plugin not installed"
+  config.hostmanager.enabled = false
+
+  config.vm.provider :virtualbox do |vb|
+    # On VirtualBox, we don't have guest additions or a functional vboxsf
+    # in CoreOS, so tell Vagrant that so it can be smarter.
+    vb.check_guest_additions = false
+    vb.functional_vboxsf     = false
   end
 
-  config.vm.box = "capgemini/apollo"
-  config.hostmanager.enabled = true
-  config.hostmanager.manage_host = true
-  config.hostmanager.include_offline = true
   config.ssh.insert_key = false
 
   # Common ansible groups.
   ansible_groups = groups['ansible_groups']
+  # We need to use a custom python interpreter for CoreOS because there is no
+  # python installed on the system.
+  ansible_groups["all:vars"] = {
+    "ansible_python_interpreter" => "\"PATH=/home/core/bin:$PATH python\""
+  }
   ansible_groups["mesos_masters"] = []
 
   masters_conf = conf['masters']
   masters_n    = masters_conf['ips'].count
   master_infos = []
 
+  # Mesos slave nodes
+  slaves_conf = conf['slaves']
+  ansible_groups["mesos_slaves"] = []
+  slave_n = slaves_conf['ips'].count
+
+  # etcd discovery token
+  total_instances = slave_n + masters_n
+  etcd_discovery_token(total_instances)
+
   # Mesos master nodes
   (1..masters_n).each { |i|
 
@@ -61,7 +83,7 @@ Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
         machine.vm.hostname = node[:hostname]
         machine.vm.network :private_network, :ip => node[:ip]
 
-        vb.name = 'vagrant-mesos-' + node[:hostname]
+        vb.name = 'coreos-mesos-' + node[:hostname]
         vb.customize ["modifyvm", :id, "--memory", node[:mem], "--cpus", node[:cpus] ]
       end
     end
@@ -83,19 +105,14 @@ Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
     consul_join: consul_join,
     consul_retry_join: consul_retry_join,
     mesos_master_quorum: conf['mesos_master_quorum'],
-    consul_bootstrap_expect: conf['consul_bootstrap_expect']
+    consul_bootstrap_expect: conf['consul_bootstrap_expect'],
+    ansible_python_interpreter: 'PATH=/home/core/bin:$PATH python'
   }
   # Apollo environment variables
   apollo_vars = get_apollo_variables(ENV)
   # Add apollo variables to ansible ones
   ansible_extra_vars.merge!(apollo_vars)
 
-  # Mesos slave nodes
-  slaves_conf = conf['slaves']
-  ansible_groups["mesos_slaves"] = []
-
-  slave_n = slaves_conf['ips'].count
-
   (1..slave_n).each { |i|
 
     ip = slaves_conf['ips'][i - 1]
@@ -109,39 +126,37 @@ Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
     # Add the node to the correct ansible group.
     ansible_groups["mesos_slaves"].push(node[:hostname])
 
+    # Bootstrap the machines for CoreOS first
+    if File.exist?(CLOUD_CONFIG_PATH)
+      config.vm.provision :file, :source => "#{CLOUD_CONFIG_PATH}", :destination => "/tmp/vagrantfile-user-data"
+      config.vm.provision :shell, :inline => "mv /tmp/vagrantfile-user-data /var/lib/coreos-vagrant/", :privileged => true
+    end
+
+    config.vm.provision :hostmanager
+
     config.vm.define node[:hostname] do |cfg|
       cfg.vm.provider :virtualbox do |vb, machine|
         machine.vm.hostname = node[:hostname]
         machine.vm.network :private_network, :ip => node[:ip]
 
-        vb.name = 'vagrant-mesos-' + node[:hostname]
+        vb.name = 'coreos-mesos-' + node[:hostname]
         vb.customize ["modifyvm", :id, "--memory", node[:mem], "--cpus", node[:cpus] ]
 
         # We invoke ansible on the last slave with ansible.limit = 'all'
         # this runs the provisioning across all masters and slaves in parallel.
         if node[:hostname] == "slave#{slave_n}"
+
           machine.vm.provision :ansible do |ansible|
             ansible.playbook = "site.yml"
-            ansible.sudo = true
             unless ENV['ANSIBLE_LOG'].nil? || ENV['ANSIBLE_LOG'].empty?
-              ansible.verbose = "#{ENV['ANSIBLE_LOG'].delete('-')}"
+             ansible.verbose = "#{ENV['ANSIBLE_LOG'].delete('-')}"
             end
-            ansible.groups = ansible_groups
-            ansible.limit = 'all'
+            ansible.groups     = ansible_groups
+            ansible.limit      = 'all'
             ansible.extra_vars = ansible_extra_vars
           end
         end
       end
     end
   }
-
-  # If you want to use a custom `.dockercfg` file simply place it
-  # in this directory.
-  if File.exist?(".dockercfg")
-    config.vm.provision :shell, :priviledged => true, :inline => <<-SCRIPT
-      cp /vagrant/.dockercfg /root/.dockercfg
-      chmod 600 /root/.dockercfg
-      chown root /root/.dockercfg
-    SCRIPT
-  end
 end

ansible-groups.yml

@@ -18,4 +18,4 @@ ansible_groups:
   "vagrant:children":
     - mesos_masters
     - mesos_slaves
-    - load_balancers
+    - load_balancers

ansible.cfg

@@ -4,6 +4,7 @@ record_host_keys = no
 jinja2_extensions = jinja2.ext.do
 timeout = 15
 gathering = smart
+roles_path = roles
 
 [ssh_connection]
 ssh_args = -o ControlMaster=auto -o ControlPersist=30m

group_vars/all

@@ -35,4 +35,4 @@ datadog_config:
   tags: "{{ mesos_cluster_name }}"
   log_level: INFO
 
-
+coreos_timezone: 'Europe/London'

playbooks/coreos-bootstrap.yml

@@ -0,0 +1,14 @@
+- name: bootstrap coreos hosts
+  hosts: all
+  gather_facts: False
+  roles:
+    - coreos_bootstrap
+    - coreos_timezone
+
+- name: Install docker-py
+  hosts: all
+  gather_facts: False
+  tasks:
+    - pip:
+        name: docker-py
+        version: 1.5.0

requirements.yml

@@ -0,0 +1,5 @@
+- src: mkaag.coreos-timezone
+  name: coreos_timezone
+
+- src: defunctzombie.coreos-bootstrap
+  name: coreos_bootstrap

roles/cadvisor/tasks/main.yml

@@ -1,35 +1,10 @@
 # tasks for running cadvisor
-- name: destroy old cadvisor container
-  when: cadvisor_rebuild_container|bool
-  docker:
-    name: cadvisor
-    image: "{{ cadvisor_image }}"
-    state: absent
-  tags:
-    - cadvisor
-
-- name: run cadvisor container
-  docker:
-    name: cadvisor
-    image: "{{ cadvisor_image }}"
-    state: started
-    restart_policy: "{{ cadvisor_restart_policy }}"
-    net: "{{ cadvisor_net }}"
-    hostname: "{{ cadvisor_hostname }}"
-    volumes:
-    - "/var/lib/docker/:/var/lib/docker:ro"
-    - "/:/rootfs:ro"
-    - "/var/run:/var/run:rw"
-    - "/sys:/sys:ro"
-  tags:
-    - cadvisor
-
-- name: upload cadvisor template service
-  template:
-    src: cadvisor.conf.j2
-    dest: /etc/init/cadvisor.conf
-    mode: 0755
+- name: deploy cadvisor service
   sudo: yes
+  sudo_user: root
+  template:
+    src: cadvisor.service.j2
+    dest: /etc/systemd/system/cadvisor.service
   tags:
     - cadvisor
 
@@ -53,13 +28,13 @@
   tags:
     - cadvisor
 
-- name: Set cadvisor consul service definition
-  sudo: yes
-  template:
-    src: cadvisor-consul.j2
-    dest: "{{ cadvisor_consul_dir }}/cadvisor.json"
-  notify:
-    - restart consul
-  when: cadvisor_enabled
-  tags:
-    - cadvisor
+#- name: Set cadvisor consul service definition
+#  sudo: yes
+#  template:
+#   src: cadvisor-consul.j2
+#    dest: "{{ cadvisor_consul_dir }}/cadvisor.json"
+#  notify:
+#    - restart consul
+#  when: cadvisor_enabled
+#  tags:
+#    - cadvisor

roles/cadvisor/templates/cadvisor.service.j2

@@ -0,0 +1,27 @@
+[Unit]
+Description=cadvisor
+After=docker.service
+Requires=docker.service
+
+[Service]
+Restart=on-failure
+RestartSec=20
+TimeoutStartSec=0
+EnvironmentFile=-/etc/environment
+ExecStartPre=-/usr/bin/docker kill cadvisor
+ExecStartPre=-/usr/bin/docker rm cadvisor
+ExecStartPre=/usr/bin/docker pull {{ cadvisor_image }}
+ExecStart=/usr/bin/docker run --name cadvisor \
+--restart={{ cadvisor_restart_policy }} \
+--net={{ cadvisor_net }} \
+--hostname={{ cadvisor_hostname }} \
+-v /var/lib/docker/:/var/lib/docker:ro \
+-v /:/rootfs:ro \
+-v /var/run:/var/run:rw \
+-v /sys:/sys:ro \
+{{ cadvisor_image }}
+
+ExecStop=/usr/bin/docker stop cadvisor
+
+[Install]
+WantedBy=multi-user.target

roles/consul/defaults/main.yml

@@ -2,10 +2,20 @@
 # defaults file for consul
 consul_dc: dc1
 consul_servers_group: consul_servers
+consul_bootstrap_expect: "{{ groups[consul_servers_group] | length }}"
 consul_advertise: "{{ ansible_ssh_host }}"
+consul_config_dir: /etc/consul.d
+consul_data_dir: /var/lib/consul
+consul_atlas_join: false
 consul_bind_addr: "{{ ansible_default_ipv4.address }}"
 consul_retry_join: "{% for host in groups[consul_servers_group] %}\"{{ hostvars[host].ansible_default_ipv4.address }}\"{% if not loop.last %}, {% endif %}{% endfor %}"
-consul_bootstrap_expect: "{{ groups[consul_servers_group] | length }}"
 consul_client_addr: "0.0.0.0"
-consul_atlas_join: false
 consul_node_name: "{{ ansible_hostname }}"
+consul_version: 0.6
+consul_image: "
+  {%- if inventory_hostname in groups[consul_servers_group] -%}
+    gliderlabs/consul-server:{{ consul_version }}
+  {%- else -%}
+    gliderlabs/consul-agent:{{ consul_version }}
+  {%- endif -%}
+"

roles/consul/handlers/main.yml

@@ -1,5 +1,6 @@
 ---
 # handlers file for consul
+# @todo - handle restarts properly (choose one approach here)
 - name: restart consul
   service:
     name: consul
@@ -12,3 +13,7 @@
   wait_for:
     host: "{{ consul_bind_addr }}"
     port: 8500
+
+- name: restart consul systemd
+  sudo: yes
+  command: systemctl restart consul

roles/consul/meta/main.yml

@@ -69,7 +69,7 @@ galaxy_info:
   #  - 9.1
   #  - 9.1
   #  - 9.2
-  - name: Ubuntu
+  - name: CoreOS
     versions:
   #  - all
   #  - lucid
@@ -80,7 +80,7 @@ galaxy_info:
   #  - quantal
   #  - raring
   #  - saucy
-     - trusty
+  #   - trusty
   #- name: SLES
   #  versions:
   #  - all

roles/consul/tasks/config.yml

@@ -0,0 +1,22 @@
+- name: create consul dirs
+  sudo: yes
+  file:
+    path: "{{ item }}"
+    state: directory
+    mode: 0755
+  with_items:
+    - "{{ consul_data_dir }}"
+    - "{{ consul_config_dir }}"
+
+- name: configure consul
+  sudo: yes
+  template:
+    src: consul.json.j2
+    dest: /etc/consul.d/consul.json
+    owner: root
+    group: root
+    mode: 0644
+  notify:
+    - restart consul
+  tags:
+    - consul