V2 Deployment Checklist

V2Master | V2Deployment

Introduction

DNS Name Configuration

  1. DNS configuration is performed on the Puppet server mgt.vh01.softxs.ch
    • ssh -p 10104 root@loki
      cd /etc/puppet
  2. Configure new DNS names. As root on the Puppet server

    • cd /etc/puppet/modules/bind9/files/master
      
      # -- Edit the following two files:
      
      vi zones/all/{domain_name}.
      # Update the 'Serial' with today's date and the sequence number (normally '01')
      :x
      
      vi generic/all/{domain_name}.
      # Add CNAME record for the sub-domain pointing to the host server
      :x
  3. Test DNS Names. Make sure the new DNS name is defined on the all following DNS servers:
    1. zg-1.softxs.ch - Public server

    2. zg-3.softxs.ch - Public server

    3. loki.softxs.ch - LU DMZ

    4. odin.softxs.ch - LU DMZ

    5. modi.softxs.ch - LU Internal network

Setup Virtual Machine & Virtual Host

Requires the following steps:

  1. Define VM Vagrantfile generation
  2. Define VM configuration
  3. Define Virtual Host configuration
  4. Start new VM
  5. Log into VM and perform the initial OS update and first puppet agent run

  6. Setup SSH login certificate

Configuration for Vagrantfile Generation

  1. Configure the Vagrantfile generation in the Puppet server mgt.vh01.softxs.ch. As user root on the Puppet server

    • ssh -p 10104 root@lu-4.softxs.ch # loki.softxs.ch
      cd /etc/puppet/modules/manifests
      
      vi nodes.pp
      
      # Find the node definition for the host server (zg-1, loki, odin, etc.)
      # Add a ''v2_server::vm'' definition, similar to the following:
      
      node "zg-3.softxs.ch" { # host VM
        ...
        create_resources( v2_server::vm, { # VM definition
            ...
            'v0402' => {
                vm_hostname   => 'v0402',
                vm_fqdn       => 'v0402.vh03.softxs.ch',
                vm_box        => 'precise64_rails',
                vm_memory     => 3072,
                vm_ip         => '172.16.4.2',
                vm_netmask    => '255.240.0.0',
                vm_ssh_port   => 20402,
                vm_synched_folders => {
                  'vagrant'  => { vm_dir => "/vagrant",       host_dir => "." },
                  'data'     => { vm_dir => "/data", host_dir => "/v01/data/v0402",
                                  owner => "www-data", group => "v2", mount_options => ["dmode=775", "fmode=764"],
                  },
                  'backup'   => { vm_dir => "/backup", host_dir => "/v01/backup/local/v0402",
                                  owner => "www-data", group => "v2", mount_options => ["dmode=775", "fmode=764"],
                  },
                },
            },
        ...
  2. The critical parameters are: (See V2VirtualServers for example configurations and the setup conventions)

    1. Hostname
    2. Fully qualified domain name (fqdn)
    3. Virtual machine box

    4. Memory
    5. IP address
    6. Port number
    7. Paths for the /data & /backup directories

Define Reverse Proxy

  1. Define a reverse proxy entry on the host machine, to forward http requests to the VM. As user root on the Puppet server

    • cd /etc/puppet
      vi manifests/nodes.pp
      
      node "zg-3.softxs.ch" { # Find the node of the host machine
        ...
        create_resources( rproxy::vhost, {
          ...
          # Add an entry similar to the following:
          'v0402' => { host => 'v030402.softxs.ch', ip => '172.16.4.2' },
          ...
        }
        ...
      }
    • Where:
      1. v0402 is the hostname of the VM

      2. host is the external (public) DNS name

      3. ip is the internal IP address of the VM

Define Application Backup & Data Directories

  1. Define Backup & Data directories for the VM. As user root on the Puppet server

    • cd /etc/puppet
      vi manifests/nodes.pp
      
      node "zg-3.softxs.ch" { # Find the node of the host machine
        ...
        create_resources( v2_server::mkdir, {
          ...
          # Add entries similar to the following
          "v0402-backup"   => { target => "/v01/backup/local/v0402" },
          "v0402-data"     => { target => "/v01/data/v0402" },
          ...
        }
        ...
      }
    • Where:
      1. v0402 is the hostname of the VM

      2. /v01/backup/local is the top-level directory of host machine's local backup tree

      3. /v01/data is the top-level directory of host machine's application directory tree

Run Puppet Agent on the Host Server

  1. Wait for puppet agent to run on the host machine to implement the updates

    1. Or run it by hand:
      • puppet agent --onetime --no-daemonize --verbose
    2. This will also create a new VM directory on the host machine:
      • /home/vagrant/vms/{name}

        Where name is the VM new host name

Document New Virtual Machine on Wiki

  1. Document the new VM in the Wiki page: V2VirtualServers

Define a Puppet Node for the New Virtual Machine

  1. Define a new puppet node for the vm, named with the VM's fully qualified domain name:

    • node "v0402.vh03.softxs.ch" { # VM FQDN
        class { 'puppet_agent': period => 10 }
        class { 'postfix_satellite': node_name => $name, mailname => 'softxs.ch', relayhost => 'smtp.softxs.ch' }
        include apache
        include mysql
        include passenger
        class { 'rails_user': user => 'v2' }
      
        # ..place for vhost definition (see next section)..
      }

Define a Virtual Host on the VM

  1. Add an apache::vhost definition to the VM node:

    • node "v0402.vh03.softxs.ch" { # VM node
        ...
        apache::vhost {'trial.softxs.ch':
          template => 'apache/vhost-v2.conf.erb',
          docroot => '/home/v2/rails/trial.softxs.ch',
          priority => 25,
          servername => 'trial.softxs.ch',
          options => '-Indexes',
          subvhosts => {
            'trial1' => { },
            'trial2' => { },
        },
        ...
      }
  2. The subvhosts are used to define the relative paths for the individual applications

Start Virtual Machine

  1. As user vagrant on the host machine start the VM:

    • cd /home/vagrant/vms/{name} # Where 'name' is the new VM host name
      vagrant up # Starts the VM
  2. Log into the VM
    • vagrant ssh
  3. Perform an initial software update on the VM
    • su -
      apt-get update
  4. Perform an initial puppet agent run on the VM

    • puppet agent --onetime --no-daemonize --verbose
    • This should create a vhost, including:
      1. Apache configuration files, located at:
        1. /etc/apache2/sites-available
        2. /etc/apache2/sites-enabled
      2. Base directories for the application(s):
        1. /home/v2/{domain}/{app(s)}

Setup V2 User SSH Certificate

  1. In order to allow automatic login for application deployment using Capistrano, you must install a SSH key on the VM for the v2 user:
    • ssh -p {port} v2@{host}
    • Where
      • port is the SSH port number defined in the VM's Vagrantfile

      • host is the base servers domain name

  2. Create a .ssh directory. As user v2 on the VM:

    • cd ~
      mkdir .ssh # If the directory doesn't already exist
      chmod 0700 .ssh
      
      cd .ssh
      vi authorized_keys
      # Add your SSH public key, typically from your local machine, in the file ''~/.ssh/identity.pub'' or ''~/.ssh/id_dsa.pub''
      :x
      
      chmod 0600 authorized_keys
  3. Test the key. As your local user on your development (or other) system, from which you plan to do application deployment, make sure the following command works without having to enter a password:
    • ssh -p {port} v2@host
    • Where:
      • port is the SSH port number defined in the VM's Vagrantfile

      • host is the base servers domain name

Setup V2 Deployment Instance

  1. Work Organiser application is done from the git repository: git.softxs.ch:/home/git/gitroot/v2deploy.git

Define Instance

  1. Go to the Works Organiser instance directory: At the top of v2deploy repository:

    • cd app/v2p0/site/proto/instance
  2. Create an new instance directory. Typically it's named {dns_name}-{path}, where
    1. dns_name is the DNS name used to access the application host (a VM)

    2. path is the relative path

    3. For example: demo.softxs.ch-hydro, which would be access with the url:

  3. Create the following directory & files in the instance directory:

    1. deploy.rb - Capistrano deployment paramenters, including server, directories, etc

    2. files - directory

      1. config - directory

        1. database.yml - Define the application's database

        2. email_secret.txt - Key for sending email

        3. settings.local.yml - Define the site/instance application specific settings

    3. The settings for these files are explained in the following sections

Define deploy.rb

  1. The deploy.rb file defines the Capistrano deployment parameters.

  2. Example deploy.rb file:

    • # deploy.rb
      # app: v2p0, site: proto, instance: demo.softxs.ch-tasks
      
      Capistrano::Configuration.instance.load do
      
        # -- Server parameters
      
        role :web,       "demo.softxs.ch"
        role :app,       "demo.softxs.ch"
        role :db,        "demo.softxs.ch", :primary => true
      
        set :vm_ssh_port, 20108
      
        set :basepath,  "/home/v2/rails"
        set :servpath,  "demo.softxs.ch"
        set :relpath,   "tasks"
      
        set :deploy_to,  "#{basepath}/#{servpath}/#{relpath}-app"
      
        # -- Relative path on server (disable if no relative path)
        # WARNING: RAILS_RELATIVE_URL_ROOT must = '/' + Settings.SXS.Application.Url.Path
        set :asset_env,  "RAILS_RELATIVE_URL_ROOT='/#{relpath}'"
      
        # -- Owner/group of deployed code
      
        set :user,       "v2"         #
        set :group,      "www-data"   #
      
        # -- Git parameters
      
        set :branch,     "master"     # Git branch to deploy
        set :scm_user,   ENV['CAP_USER'] || ENV['USER']
      
        # -- Installation
      
        set :use_sudo,   false
      end
  3. The key settings are:
    1. roles - The server(s) where the application will be defined. Normally they are all set to the same machine

      1. :web - The web server

      2. :app - The application server

      3. :db - The SQL server

    2. :vm_ssh_port - The SSH port to use to log into the server(s)

    3. :basepath - The absolute path where applications are installed

    4. :servpath - The relative path to the vhost

    5. :relpath - The relative path to the application

    6. :branch - The git branch to install

settings.local.yml

database.yml

Email Secret Key

Home Page Setup

Deploy Initial V2 System

Capistrano Deployment

Test Local Login

Test Invitation & MAPS Login

Start Delayed Job Task

Register Application in MAPS

Test Outgoing Email

Configure & Test Incoming Email

Install V2 Configuration

Import Configuration

Import Data & Documents

End-to-End Test

Setup Backups

Application backups run on a nightly basis. There are three parts:

  1. A database backup to the application's ../shared/backup directory

  2. A backup of the application's -app directory tree (which includes shared/backup) to a sub-directory in the MV's /backup (NFS mount-ed from the host server)

  3. A backup server rsync-ing the host machine's backup area (TODO: described elsewhere)

Backups are based on the following cronjobs:

  1. On the application VM, setup cronjobs as user v2:

    1. Backup the application database
    2. Backup the application's -app directory tree

Application Database Backup

  1. On the application VM, as user v2 Setup a cronjob like the following:

    • 23 03 * * * /home/v2/rails/{domain}/{relpath}-app/current/script/ruby_cron.sh \
                    /home/v2/rails/{domain}/{relpath}-app/current/script/backup_db \
                      -o /home/v2/rails/{domain}/{relpath)-app/shared/backup \
                      -u{usr} -p{pwd} -z
    • The cronjob should run every 24 hours and its time should be before the root rsync cronjob shown below

    • Where:
      1. {domain} is the application's domain name. E.g. poyry.works-organiser.com

      2. {relpath} is the relative URL path to the application. E.g. stadelhofen

      3. {usr} is the MySQL administrative user's name. E.g. root

      4. {pwd} is the MySQL administrative user's name. E.g. secret

      5. -z is the option to compress the output with bzip2

Application Directory Tree Backup

Warning: These cronjobs currently run as root, as there are permissions errors running as user v2

  1. On the application VM, as user v2 Setup a cronjob like the following:

    • /usr/bin/rsync -qa /home/v2/rails/{domain}/{relpath}-app/ /backup/{domain}/{relpath}/
    • The cronjob should run every 24 hours and its time should be before the backup server's collection of the host machine's nightly backup data

    • Where:
      1. {domain} is the application's domain name. E.g. poyry.works-organiser.com

      2. {relpath} is the relative URL path to the application. E.g. stadelhofen

Setup Monitoring

V2DeploymentChecklist (last edited 2015-05-12 08:55:54 by 172)

Copyright 2008-2014, SoftXS GmbH, Switzerland