cookbook 'ark', '= 1.0.1', :supermarket
Provides a custom resource for installing runtime artifacts in a predictable fashion
cookbook 'ark', '= 1.0.1'
knife supermarket install ark
knife supermarket download ark
This cookbook provides
ark, a resource for managing software archives. It manages the fetch-unpack-configure-build-install process common to installing software from source, or from binary distributions that are not fully fledged OS packages.
This cookbook started its life as a modified version of Infochimp's install_from cookbook. It has since been heavily refactored and extended to meet different use cases.
Given a simple project archive available at a url:
ark 'pig' do url 'http://apache.org/pig/pig-0.8.0.tar.gz' end
The provider will:
- fetch it to to
- unpack it to the default path (
- create a symlink for
/usr/local/pig) pointing to path
- add specified binary commands to the environment
By default, the ark will not run again if the
:path is not empty. Ark provides many actions to accommodate different use cases, such as
At this time ark only handles files available from URLs using the remote_file provider. It does handle local files using the
- Mac OS X
Should work on common Unix/Linux systems with typical userland utilities like tar, gzip, etc. May require the installation of build tools for compiling from source, but that installation is outside the scope of this cookbook.
- Chef 11+
Customize the attributes to suit site specific conventions and defaults.
node['ark']['apache_mirror'] - if the URL is an apache mirror, use the attribute as the default.
node['ark']['prefix_root'] - default base location if the
prefix_root is not passed into the resource.
node['ark']['prefix_bin'] - default binary location if the
prefix_bin is not passed into the resource.
node['ark']['prefix_home'] - default home location if the
prefix_home is not passed into the resource.
node['ark']['win_install_dir'] - directory where the files will be put on windows
node['ark']['package_dependencies'] - prerequisite system packages that need to be installed to support ark.
ark- does the extract/build/configure dance
:install: extracts the file and creates a 'friendly' symbolic link to the extracted directory path
:configure: configure ahead of the install action
:install_with_make: extracts the archive to a path, runs
make install. It does not run the configure step at this time
:dump: strips all directories from the archive and dumps the contained files into a specified path
:cherry_pick: extract a specified file from an archive and places in specified path
:put: extract the archive to a specified path, does not create any symbolic links
:remove: removes the extracted directory and related symlink #TODO
:setup_py: runs the command "python setup.py" in the extracted directory
:setup_py_build: runs the command "python setup.py build" in the extracted directory
:setup_py_install: runs the command "python setup.py install" in the extracted directory
Extract a specified file from an archive and places in specified path.
Relevant Attribute Parameters for :cherry_pick
path: directory to place file in.
creates: specific file to cherry-pick.
Strips all directories from the archive and dumps the contained files into a specified path.
NOTE: This currently only works for zip archives
Attribute Parameters for :dump
path: path to dump files to.
mode: file mode for
app_home, as an integer.
creates: if you are appending files to a given directory, ark needs a condition to test whether the file has already been extracted. You can specify with creates, a file whose existence indicates the ark has previously been extracted and does not need to be extracted again.
Extract the archive to a specified path, does not create any symbolic links.
Attribute Parameters for :put
path: path to extract to.
append_env_path: boolean, if true, append the
./bindirectory of the extracted directory to the global
PATHvariable for all users.
name: name of the package, defaults to the resource name.
url: url for tarball,
.zipcurrently supported. Also supports special syntax
:name:version:apache_mirror:that will auto-magically construct download url from the apache mirrors site.
version: software version, defaults to
mode: file mode for
app_home, is an integer.
prefix_root, for use with
prefix_home: default directory prefix for a friendly symlink to the path.
prefix_bin: default directory to place a symlink to a binary command.
/opt/maven-2.2.1/bin/mvn, where the
path: path to extract the ark to. The
:install*actions overwrite any user-provided values for
home_dir: symbolic link to the path
:prefix_root/:name-:version, does not apply to
has_binaries: array of binary commands to symlink into
/usr/local/bin/, you must specify the relative path.
[ 'bin/java', 'bin/javaws' ]
append_env_path: boolean, similar to
has_binariesbut less granular. If true, append the
./bindirectory of the extracted directory to. the
PATHenvironment variable for all users, by placing a file in
/etc/profile.d/. The commands are symbolically linked into
/usr/bin/*. This option provides more granularity than the boolean option.
environment: hash of environment variables to pass to invoked shell commands like
strip_components: number of components in path to strip when extracting archive. With default value of
1, ark strips the leading directory from an archive, which is the default for both
autoconf_opts: an array of command line options for use with the GNU
[ '--include=/opt/local/include', '--force' ]
make_opts: an array of command line options for use with
[ '--warn-undefined-variables', '--load-average=2' ]
owner: owner of extracted directory.
This example copies
/var/cache/chef/ivy-2.2.0.tar.gz, unpacks its contents to
/usr/local/ivy-2.2.0/ -- stripping the leading directory, and symlinks
# install Apache Ivy dependency resolution tool ark "ivy" do url 'http://someurl.example.com/ivy.tar.gz' version '2.2.0' checksum '89ba5fde0c596db388c3bbd265b63007a9cc3df3a8e6d79a46780c1a39408cb5' action :install end
This example copies
/var/cache/chef/jdk-7.2.tar.gz, unpacks its contents to
/usr/local/jvm/jdk-7.2/ -- stripping the leading directory, symlinks
/usr/local/jvm/jdk-7.2, and adds
/usr/local/jvm/jdk-7.2/bin/ to the global
PATH for all users. The user 'foobar' is the owner of the
ark 'jdk' do url 'http://download.example.com/jdk-7u2-linux-x64.tar.gz' version '7.2' path "/usr/local/jvm/" home_dir "/usr/local/jvm/default" checksum '89ba5fde0c596db388c3bbd265b63007a9cc3df3a8e6d79a46780c1a39408cb5' append_env_path true owner 'foobar' end
Install Apache Ivy dependency resolution tool in <path>/resource_name in this case
/usr/local/ivy, do not symlink, and strip any leading directory if one exists in the tarball:
ark "ivy" do url 'http://someurl.example.com/ivy.tar.gz' checksum '89ba5fde0c596db388c3bbd265b63007a9cc3df3a8e6d79a46780c1a39408cb5' action :put end
Install Apache Ivy dependency resolution tool in /home/foobar/ivy, strip any leading directory if one exists:
ark "ivy" do path "/home/foobar" url 'http://someurl.example.com/ivy.tar.gz' checksum '89ba5fde0c596db388c3bbd265b63007a9cc3df3a8e6d79a46780c1a39408cb5' action :put end
Strip all directories and dump files into path specified by the path attribute. You must specify the
creates attribute in order to keep the extraction from running every time. The directory path will be created if it doesn't already exist:
ark "my_jars" do url "http://example.com/bunch_of_jars.zip" path "/usr/local/tomcat/lib" creates "mysql.jar" owner "tomcat" action :dump end
Extract specific files from a tarball (currently only handles one named file):
ark 'mysql-connector-java' do url 'http://oracle.com/mysql-connector.zip' creates 'mysql-connector-java-5.0.8-bin.jar' path '/usr/local/tomcat/lib' action :cherry_pick end
Build and install haproxy and use alternative values for
ark "haproxy" do url "http://haproxy.1wt.eu/download/1.5/src/snapshot/haproxy-ss-20120403.tar.gz" version "1.5" checksum 'ba0424bf7d23b3a607ee24bbb855bb0ea347d7ffde0bec0cb12a89623cbaf911' make_opts [ 'TARGET=linux26' ] prefix_root '/opt' prefix_home '/opt' prefix_bin '/opt/bin' action :install_with_make end
You can also pass multiple actions to ark and supply the file extension in case the file extension can not be determined by the URL:
ark "test_autogen" do url 'https://github.com/zeromq/libzmq/tarball/master' extension "tar.gz" action [ :configure, :install_with_make ] end
License and Authors
- Author: Philip (flip) Kromer - Infochimps, Inc(email@example.com)
- Author: Bryan W. Berry (firstname.lastname@example.org)
- Author: Denis Barishev (email@example.com)
- Author: Sean OMeara (firstname.lastname@example.org)
- Author: John Bellone (email@example.com)
- Copyright: 2011, Philip (flip) Kromer - Infochimps, Inc
- Copyright: 2012, Bryan W. Berry
- Copyright: 2012, Denis Barishev
- Copyright: 2013, Chef Software, Inc
- Copyright: 2014, Bloomberg L.P.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
|seven_zip >= 0.0.0|
|windows >= 0.0.0|
|build-essential >= 0.0.0|
ark Cookbook CHANGELOG
This file is used to list changes made in each version of the ark cookbook.
- Remove a large number of zero byte archives that snuck into the repository
- Remove a Chef 10 compatibility check in the custom resource
- Added the pkg-config package to the debian platform family
- Added tar, xz-lzma-compat, and bzip2 packages to the RHEL and fedora platform families
- Updated FreeBSD to install gmake instead of make
- Added OS X, SmartOS, and FreeBSD to the tar path attributes to support those platforms
- Removed the has_binaries attribute from put action documentation in the readme file since this isn't supported there
- Moved the libraries module locations to no longer be under Opscode:: and broke out libraries into more logical units
- Fixed issues with spaces in Windows paths that could cause failures
- Fixed a bad attribute for the 7zip home on windows. Instead of using a node attribute use the value directly to avoid computed attribute overiding issues
- Switched from the 7-zip cookbook to seven_zip since the 7-zip cookbook is now deprecated
- Changed unzip commands to not use -u so that a newer archive can overwrite an existing directory
- Added support for actions py_setup, py_setup_install, py_setup_build
- Fixed setting home_dir attribute
- Added source_url and issues_url to the metadata.rb
- Expanded the supported platforms in metadata.rb
- Removed all references to Opscode
- Improved error logging when an unknown extension is encountered
- Added support for .tar files
Improved overall testing:
- Removed the kitchen.cloud.yml file and gem dependencies
- Added integration testing in Travis with Kitchen-Docker and Travis tests now run using the nightly build of ChefDK
- Expanded platforms tested in the .kitchen.yml file
- Updated the Gemfile with the latest testing dependencies
- Added full Chefspec coverage
- Greatly expanded the ark_spec test cookbook
- Removed the original minitests
Added standard Chef .gitignore and chefignore files
Resolved a large number of rubocop warnings
Removed old Opscode contributing and testing docs
Added a cookbook version badge to the readme
Removed the Toftfile
- [COOK-3642] Add Windows support
- [COOK-4514] - Support for SLES with the Ark cookbook
- [COOK-2771] - Add support for XZ compression
- [COOK-4477] - Fix failing test suite
- [COOK-4484] - Replace strip_leading_dir attribute with more general strip_components
- [COOK-4437] - configure and install_with_make should chown after unpack
[COOK-3786] - Unable to install multiple versions of archive without duplication
- COOK-4288 - Cleanup the Kitchen
- COOK-4264 - Add node['ark']['package_dependencies'] to allow tuning packages.
- COOK-3854 - Capability with mac_os_x: '/bin/chown' - No such file or directory
- Cleaning up some style for rubucop
- Updating test harness
- COOK-3539 - Allow dumping of bz2 and gzip files
- COOK-3191 - Propogate unzip failures
- COOK-3118 - Set cookbook attribute in provider
- COOK-3055 - Use proper scope in helper module
- COOK-3054 - Fix notification resource updating
- COOK-3179 - README updates and refactor
- [COOK-3087]: Can't use ark with chef < 11
only_ifstatements in ark's
install_with_makeand configure actions are not testing for file existence correctly.
- [COOK-3067]: ark kitchen test for
cherry_pickis expecting the binary to be in the same parent folder as in the archive.
- [COOK-3048]: Ark provider contains a
ruby_blockresource without a block attribute
- [COOK-3063]: Ark cookbook
cherry_pickaction's unzip command does not close if statement
- [COOK-3065]: Ark install action does not symlink binaries correctly
- Update the README to reflect the requirement for Chef 11 to use the ark resource (
- Making this a release so it will also appear on the community site page.
- [COOK-2772]: Ark cookbook has foodcritic failures in provides/default.rb
- [COOK-2520]: Refactor ark providers to use the '
use_inline_resources' LWRP DSL feature
- [COOK-2335] - ark resource broken on Chef 11
- [COOK-2026] - Allow
cherry_pickaction to be used for directories as well as files
- [COOK-1593] - README formatting updates for better display on Community Site
- dangling "unless"
- add vagrantfile
- add foodcritic test
- travis.ci support
v0.0.10 (May 23, 2012
strip_leading_dirnot working for zip files https://github.com/bryanwb/chef-ark/issues/19
1.0.1 passed this metric
1.0.1 passed this metric