search

RiotGamesCookbooks / artifact-cookbook

Provides your cookbooks with the Artifact Deploy LWRP
Other
134 stars 91 forks source link

readme

Artifact cookbook

Provides your cookbooks with the Artifact Deploy LWRP

Requirements

Platforms

Vagrant

With Vagrant 1.1, there is no longer a Vagrant RubyGem to install. Instead, follow the instructions on the VagrantUp documentation pages.

Resources / Providers

artifact_deploy

Deploys a collection of build artifacts packaged into a tar ball. Artifacts are extracted from the package and managed in a deploy directory in the same fashion you've seen in the Opscode deploy resource or Capistrano's default deploy strategy.

Actions

Action Description Default
deploy Deploy the artifact package Yes
pre_seed Pre-seed the artifact package

Attributes

Attribute Description Type Default
artifact_name Name of the artifact package to deploy String name
artifact_location URL, S3 path, local path, or Maven identifier of the artifact package to download String
artifact_checksum The SHA256 checksum of the artifact package that is being downloaded String
deploy_to Deploy directory where releases are stored and linked String
version Version of the artifact being deployed String
owner Owner of files created and modified String
group Group of files created and modified String
environment An environment hash used by resources within the provider Hash Hash.new
symlinks A hash that maps files in the shared directory to their paths in the current release Hash Hash.new
shared_directories Directories to be created in the shared folder Array %w{ log pids }
force Forcefully deploy an artifact even if the artifact has already been deployed Boolean false
should_migrate Notify the provider if it should perform application migrations Boolean false
keep Specify a number of artifacts deployments to keep on disk Integer 2
use_symlinks Use symlinks for pointing the current release Boolean true
before_deploy A proc containing resources to be executed before the deploy process begins Proc
before_extract A proc containing resources to be executed before the artifact package is extracted Proc
after_extract A proc containing resources to be executed after the artifac package is extracted Proc
before_symlink A proc containing resources to be executed before the symlinks are created Proc
after_symlink A proc containing resources to be executed after the symlinks are created Proc
configure A proc containing resources to be executed to configure the artifact package Proc
before_migrate A proc containing resources to be executed before the migration Proc Proc
migrate A proc containing resources to be executed during the migration stage Proc
after_migrate A proc containing resources to be executed after the migration Proc Proc
restart A proc containing resources to be executed at the end of a successful deploy Proc
after_deploy A proc containing resources to be executed after the deploy process ends Proc
after_download A proc containing resources to be executed only if the artifact has been downloaded Proc
remove_top_level_directory Deletes a top level directory from the extracted zip file Boolean false
skip_manifest_check Skips the manifest check for idempotency when the version attribute is not changing Boolean false
remove_on_force Removes the current version directory contents when force is set Boolean false
nexus_configuration Accepts an object that can customize the Nexus server connection information Chef::Artifact::NexusConfiguration Chef::Artifact::NexusConfiguration.from_data_bag

Deploy Flow, the Manifest, and Procs

The deploy flow is outlined in the Artifact Deploy flow chart below.

Artifact Deploy

For a more detailed flow of what happens when we check with deploy?, see the Manifest Differences Flow chart.

The 'happy-path' of this flow is the default path when an artifact has already been deploy - there will be no need to execute many of the Procs. That being said, there are a few 'choice' paths through the flow where a Proc may affect the flow.

There are two checks in the artifact deploy flow where a manifest check is executed - at the beginning, before the before_deploy proc, and just after the configure proc (and after the migrate procs). When the latter check returns true, the restart proc will execute.

The manifest is a YAML file with a mapping of files in the deploy path to their SHA1 checksum. For example:

/srv/artifact_test/releases/2.0.68/log4j.xml: 96be5753fbf845e30b643fa04008f2c4fe6956a7
/srv/artifact_test/releases/2.0.68/readme.txt: fcb8d816b062565930f19f9bdb954f5ac43c5039
/srv/artifact_test/releases/2.0.68/my-artifact.jar: 42ad63cc883afad010573d3d8eea4e5a4011e5d4

There are numerous Procs placed throughout the flow of the artifact_deploy resource. They are meant to give the user many different ways to configure the artifact and execute resources during the flow. Some good examples include executing a resource to stop a service in the before_deploy proc, or placing configuration files in the deployed artifact during the configure proc.

Please note the before_deploy, configure, and after_deploy procs are executed on every Chef run. It is recommended that any template (or configuration changing resource calls) take place within those procs. In particular, the configure proc was added for this very purpose. Following this pattern will ensure that the templates will change, and the restart proc will execute (perhaps restarting the service the configured artifact provides in order to pick up the configuration changes).

Procs can also utilize the internal methods of the provider class, because they are evaluated inside of the instance of the provider class. For example:

artifact_deploy "artifact_test" do
  # omitted for brevity
  configure Proc.new {
    # release_path is an attr_reader on the @release_path variable
    template "#{release_path}/conf/config.properties" do
      source "config.properties.erb"
      variables(:config => config)
    end
  }
end

By default, the current folder is a symlink pointing to the current release folder. This behavior is implicitly set by the use_symlinks. If use_symlinks is false then no symlinks are used for the current directory. Instead, files are copied from the current release directory to current and a file named .symlinks (located at the deploy_to root) contains the reference of the current release. Please note that this parameter doesn't affect how symlinks behaves.

artifact_file

Downloads a file from a provided location and then verifies that the integrity of the file is intact. Artifact files from Nexus will check with the Nexus Server to verify the SHA1 of the downloaded file. Artifact files from an HTTP or S3 source will either use the provided SHA256 checksum to verify integrity or skip the check if no checksum is given.

Actions

Action Description Default
create Download the artifact file Yes

Attributes

Attribute Description Type Default
path The path to download the artifact to String name
location The location to the artifact file. Either a nexus identifier, S3 path or URL String
checksum The SHA256 checksum for verifying URL downloads. Not used when location is Nexus String
owner Owner of the downloaded file String
group Group of the downloaded file String
after_download A proc containing resources to be executed only if the artifact has been downloaded Proc
download_retries The number of times to attempt to download the file if it fails its integrity check Integer 1

Downloading files using artifact_file

In its simplest state, the artifact_file resource is a wrapper for the remote_file resource for Nexus and URL locations. The key addition is retry logic and integrity checking for the downloaded files. Below is a brief description of the logic flow for the resource:

When the logic returns true, the downloaded file is considered good and the resource will exit. When the logic above returns false, the downloaded file is considered corrupt and an attempt will be made to download the file again. The number of retries can be controlled with the download_retries attribute.

artifact_package

Downloads a file with artifact_file, then calls the package resource to install it.

Actions

Action Description Default
install Download and install the artifact file Yes

Attributes

Attribute Description Type Default
name The name of the package you'll be installing String name
location The location to the artifact file. Either a nexus identifier, S3 path or URL String
checksum The SHA256 checksum for verifying URL downloads. Not used when location is Nexus String
owner Owner of the downloaded file String
group Group of the downloaded file String
after_download A proc containing resources to be executed only if the artifact has been downloaded Proc
download_retries The number of times to attempt to download the file if it fails its integrity check Integer 1

Documentation

The RDocs for the deploy.rb provider can be found under the Top Level Namespace page for this repository.

Nexus Usage

By default, deploying an artifact from a Nexus repository requires an encrypted data bag that contains the credentials for your Nexus repository.

knife data bag create artifact _wildcard -c <your chef config> --secret-file=<your secret file>

Your data bag should look like the following:

{
  "id": "_wildcard",
  "nexus": {
    "username": "nexus_user",
    "password": "nexus_user_password",
    "url": "http://nexus.yourcompany.com:8081/nexus/",
    "repository": "your_repository"
  }
}

After your encrypted data bag is setup you can use Maven identifiers as your artifact_location attribute. A Maven identifier is shown as a colon-separated string that includes three elements - groupId:artifactId:extension - ex. "com.my.artifact:my-artifact:tgz". If many environments share the same configuration, you can provide environment specific configuration in separate data_bag items:

knife data bag create artifact production -c <your chef config> --secret-file=<your secret file>

{
  "id": "production",
  "nexus": {
    "username": "nexus_production_user",
    "password": "nexus_production_user_password",
    "url": "http://nexus.yourcompany.com:8081/nexus/",
    "repository": "your_repository"
  }
}

knife data bag create artifact development -c <your chef config> --secret-file=<your secret file>

{
  "id": "development",
  "nexus": {
    "username": "nexus_dev_user",
    "password": "nexus_dev_user_password",
    "url": "http://nexus-dev.yourcompany.com:8081/nexus/",
    "repository": "your_repository"
  }
}

To further customize your Nexus usage, you can use the new nexus_configuration attribute. To do so, create a new Chef::Artifact::NexusConfiguration object, passing it the customized parameters - url, repository, username (defaults to nil), password (defaults to nil), ssl_verify (defaults to true). Then pass that object to the artifact_deploy resource. For example:

nexus_configuration_object = Chef::Artifact::NexusConfiguration.new("http://nexus-url", "snapshots", "username", "password")

artifact_deploy "my-artifact" do
  version             "latest"
  artifact_location   "com.foo:my-artifact:tgz"
  nexus_configuration nexus_configuration_object
  deploy_to           "/opt/my-artifact"
  owner               "artifact"
  group               "artifact"
end

To access a Nexus repository anonymously, simply omit the username and password from the construction of the Chef::Artifact::NexusConfiguration object.

S3 Usage

S3 can be used as a source of an archive. The location path must be in the form s3://s3-endpoint/bucket-name/path/to/archive.tar.gz. You can provide AWS credentials in the data_bag, or if you are running on EC2 and are using IAM Instance Roles - you may omit the credentials and use the Instance Role. Alternatively, if the credentials are available on the environment they will be used from there (more information on the Environment variable keys an be found http://docs.aws.amazon.com/AWSSdkDocsRuby/latest/DeveloperGuide/ruby-dg-roles.html).

The S3 endpoints are documented here http://docs.aws.amazon.com/general/latest/gr/rande.html#s3_region.

If you want to retrieve an object from an S3 bucket in the US-Standard region - use the following format: s3://s3.amazonaws.com/bucket-name/path/to/archive.tar.gz.

If you want to retrieve and object from a bucket in the US West (Oregon) Region region s3://s3-us-west-2.amazonaws.com/bucket-name/path/to/archive.tar.gz or EU (Ireland) Region: s3://s3-eu-west-1.amazonaws.com/bucket-name/path/to/archive.tar.gz

This is example IAM policy to get an artifact from S3. Note that this policy will limit permissions to just the files contained under the deploys/ path. If you wish to keep them in the root of your bucket, just omit the deploys/ portion and put <your-s3-bucket>/*:

{
  "Statement": [
    {
      "Sid": "Stmt1357328135477",
      "Action": [
        "s3:GetObject",
        "s3:ListBucket"
      ],
      "Effect": "Allow",
      "Resource": [
        "arn:aws:s3:::<your-s3-bucket>/deploys/*",
        "arn:aws:s3:::<your-s3-bucket>"
      ]
    }
  ]
}

If you wish to provide your AWS credentials in a data_bag, the format is:

{
  "id": "_wildcard",
  "aws": {
    "access_key_id": "my_access_key",
    "secret_access_key": "my_secret_access_key"
  }
}

Your data_bag can contain both nexus and aws configuration.

Examples

Deploying a Rails application
artifact_deploy "pvpnet" do
  version "1.0.0"
  artifact_location "https://artifacts.location.riotgames.com/pvpnet-1.0.0.tar.gz"
  deploy_to "/srv/pvpnet"
  owner "riot"
  group "riot"
  environment { 'RAILS_ENV' => 'production' }
  shared_directories %w{ data log pids system vendor_bundle assets }

  before_deploy Proc.new {
    bluepill_service 'pvpnet-unicorn' do
      action :stop
    end
  }

  before_migrate Proc.new {
    template "#{shared_path}/database.yml" do
      source "database.yml.erb"
      owner node[:merlin][:owner]
      group node[:merlin][:group]
      mode "0644"
      variables(
        :environment => environment,
        :options => database_options
      )
    end

    execute "bundle install --local --path=vendor/bundle --without test development cucumber --binstubs" do
      environment { 'RAILS_ENV' => 'production' }
      user "riot"
      group "riot"
    end
  }

  migrate Proc.new {
    execute "bundle exec rake db:migrate" do
      cwd release_path
      environment { 'RAILS_ENV' => 'production' }
      user "riot"
      group "riot"
    end
  }

  after_migrate Proc.new {
    ruby_block "remove_run_migrations" do
      block do
        Chef::Log.info("Migrations were run, removing role[pvpnet_run_migrations]")
        node.run_list.remove("role[pvpnet_run_migrations]")
      end
    end
  }

  configure Proc.new {
    template "/srv/pvpnet/current/config.properties" do
      source "config.properties.erb"
      owner 'riot'
      group 'riot'
      variables(:database_config => node[:pvpnet_cookbook][:database_config])
    end
  }

  restart Proc.new {
    bluepill_service 'pvpnet-unicorn' do
      action :restart
    end
  }

  keep 2
  should_migrate (node[:pvpnet][:should_migrate] ? true : false)
  force (node[:pvpnet][:force_deploy] ? true : false)
  action :deploy
end
Deploying the latest from Nexus (Changed in > 1.0.0)
artifact_deploy "my-artifact" do
  version           "latest"
  artifact_location "com.foo:my-artifact:tgz"
  deploy_to         "/opt/my-artifact"
  owner             "artifact"
  group             "artifact"

  before_extract Proc.new {
    service "my-artifact" do
      action :stop
    end
  }

  configure Proc.new {
    template "#{release_path}/conf/config.properties" do
      source "config.properties.erb"
      owner  "artifact"
      group  "artifact"
      variables(:config => node[:my_artifact_cookbook][:config])
    end
  }

  restart Proc.new {
    service "my-artifact" do
      action :start
    end
  }
end
Deploying an artifact from Amazon S3
artifact_deploy "my-artifact" do
  version           "1.0.0"
  artifact_location "s3://s3.amazonaws.com/my-website-deployments/deploys/my-artifact-1.0.0.tgz"
  deploy_to         "/srv/my-artifact"
  owner             node[:artifact_owner]
  group             node[:artifact_group]
  symlinks({
    "log" => "log"
  })
end
Configuring an artifact_deploy that may need to change over many Chef runs
artifact_deploy "my-artifact" do
  version           "1.0.0"
  artifact_location "http://www.fooo.com/my-artifact-1.0.0.tgz"
  deploy_to         "/srv/my-artifact"
  owner             node[:artifact_owner]
  group             node[:artifact_group]
  symlinks({
    "log" => "log"
  })
  force             node[:force_deploy]
end
Using artifact_file to download a file from a URL
artifact_file "/tmp/my-artifact.tgz" do
  location "http://www.my-website.com/my-artifact-1.0.0.tgz"
  owner "me"
  group "mes"
  action :create
end

Using artifact_file to download a file from Nexus

artifact_file "/tmp/my-artifact.tgz" do
  location "com.test:my-artifact:tgz:1.0.0"
  owner "me"
  group "mes"
  action :create
end

Configuring your resource in this manner will allow you to ensure it can always change when you need it to. In other words, configuring the force attribute to a node attribute, will allow you to change some of the more finer grained aspects of the resource. For example, when force is true, you can also change the value of owner and group to remap the deployed artifact to a new permissions scheme.

Using artifact_file to download a file from an S3 bucket
artifact_file "/tmp/my-artifact.tgz" do
  location "s3://s3.amazonaws.com/my-website-deployments/deploys/my-artifact-1.0.0.tgz"
  owner "me"
  group "mes"
  checksum "fcb188ed37d41ff2cbf1a52d3a11bfde666e036b5c7ada1496dc1d53dd6ed5dd"
  action :create
end
Using artifact_package to install an rpm from a website
artifact_package "tomcat" do
  location "http://my-website/tomcat-5.0.rpm"
  owner "me"
  group "mes"
  action :install
end
Using artifact_package to install an rpm from Nexus
artifact_package "tomcat" do
  location "com.rpm:tomcat:rpm:6.0.0"
  owner "me"
  group "mes"
  action :install
end

Releasing

Use stove

  1. Edit your ~/.stove
  2. bake ${VERSION}

License and Author

Author:: Jamie Winsor (jamie@vialstudios.com)
Author:: Kyle Allan (kallan@riotgames.com)

Copyright 2013, Riot Games

See LICENSE for license details