/cookbook-cq

Chef cookbook for Adobe CQ (aka AEM)

Primary LanguageRubyOtherNOASSERTION

CQ/AEM Chef cookbook

This cookbook deploys and configures Adobe Experience Manager (AEM), formerly known as CQ.

FYI, it is not called aem-coobkook, because when I started development there was no AEM yet (it was known as CQ at that time). Nowadays the aem name seems to be taken anyways, so I no longer have a choice.

Table of contents

Supported platforms

Operating systems

  • CentOS/RHEL 6.x
  • CentOS/RHEL 7.x
  • CentOS/RHEL 8.x
  • Amazon Linux

Chef versions


IMPORTANT

Since version 2.0.0 only Chef Infra Client 16.x is supported. Please stick to version 1.x if you're using older version of Chef!


  • Chef 16.x

AEM/CQ versions

  • AEM 6.1.0
  • AEM 6.2.0
  • AEM 6.3.0
  • AEM 6.4.0
  • AEM 6.5.0

Getting started

TBD

Attributes

For default values please refer to appropriate files.

default.rb

To set Java related attributes please refer to java cookbook. By default it installs Oracle's JDK7.

  • ( String ) node['cq']['user'] - System user for CQ/AEM service
  • ( String ) node['cq']['user_uid'] - UID of CQ/AEM user
  • ( String ) node['cq']['user_comment'] - Comment/description of CQ/AEM user
  • ( String ) node['cq']['user_shell'] - Default shell of CQ/AEM user
  • ( String ) node['cq']['group'] - System group for CQ/AEM
  • ( String ) node['cq']['group_gid'] - GID of CQ/AEM group
  • ( String ) node['cq']['limits']['file_descriptors'] - Max number of open file descriptor for CQ/AEM user
  • ( String ) node['cq']['base_dir'] - Base directory for CQ/AEM instance(s)
  • ( String ) node['cq']['home_dir'] - Home directory under which CQ/AEM instances are deployed
  • ( String ) node['cq']['version'] - CQ/AEM version
  • ( String ) node['cq']['custom_tmp_dir'] - Custom directory that JVM uses for temporary files
  • ( String ) node['cq']['jar']['url'] - URL from which CQ/AEM JAR file is downloaded
  • ( String ) node['cq']['jar']['checksum'] - SHA256 checksum of CQ/AEM JAR file
  • ( String ) node['cq']['license']['url'] - URL from which CQ/AEM license is downloaded
  • ( String ) node['cq']['license']['checksum'] - SHA256 checksum of CQ/AEM license file
  • ( Integer ) node['cq']['service']['start_timeout'] - Max number of seconds to wait until CQ/AEM instance is fully operational after service start
  • ( Integer ) node['cq']['service']['kill_delay'] - Max number of seconds for graceful instance stop before KILL signal is sent to the process
  • ( Integer ) node['cq']['service']['restart_sleep'] - Number of seconds to wait between service stop and start
  • ( String ) node['cq']['init_template_cookbook'] - Cookbook which is a source for init script template
  • ( String ) node['cq']['conf_template_cookbook'] - Cookbook which is a source for conf file template

author.rb

All attributes in this file refer to CQ/AEM author instance (node['cq']['author'] namespace).

  • ( String ) node['cq']['author']['run_mode'] - Instance run mode
  • ( String ) node['cq']['author']['port'] - Main port of CQ/AEM instance
  • ( String ) node['cq']['author']['jmx_ip'] - Value of -Djava.rmi.server.hostname JVM parameter. Requires reference to ${CQ_JMX_IP} shell variable in node['cq']['author']['jvm']['jmx_opts'] attribute to be effective
  • ( String ) node['cq']['author']['jmx_port'] - Value of -Dcom.sun.management.jmxremote.port and -Dcom.sun.management.jmxremote.rmi.port JVM parameters. Requires reference to ${CQ_JMX_PORT} shell variable in node['cq']['author']['jvm']['jmx_opts'] attribute to be effective
  • ( String ) node['cq']['author']['debug_ip'] - IP to listen on with debug interface. Requires reference to ${CQ_DEBUG_IP} shell variable in node['cq']['author']['jvm']['debug_opts'] attribute to be effective
  • ( String ) node['cq']['author']['debug_port'] - Port of JVM debug interface. Requires reference to ${CQ_DEBUG_PORT} shell variable in node['cq']['author']['jvm']['debug_opts'] attribute to be effective
  • ( String ) node['cq']['author']['credentials']['login'] - User that's used to perform actions on your CQ/AEM instance. The most typical scenarios require admin
  • ( String ) node['cq']['author']['credentials']['password'] - Password of user specified in node['cq']['author']['credentials']['login']
  • ( String ) node['cq']['author']['jvm']['min_heap'] - -Xms JVM parameter (in megabytes)
  • ( String ) node['cq']['author']['jvm']['max_heap'] - -Xmx JVM parameter (in megabytes)
  • ( String ) node['cq']['author']['jvm']['max_perm_size'] - -XX:MaxPermSize JVM parameter (in megabytes)
  • ( String ) node['cq']['author']['jvm']['code_cache_size'] - -XX:ReservedCodeCacheSize JVM parameter (in megabytes)
  • ( String ) node['cq']['author']['jvm']['general_opts'] - Generic JVM parameters
  • ( String ) node['cq']['author']['jvm']['code_cache_opts'] - JVM parameters related to its code cache
  • ( String ) node['cq']['author']['jvm']['gc_opts'] - JVM parameters related to garbage collection
  • ( String ) node['cq']['author']['jvm']['jmx_opts'] - JVM parameters related to JMX settings
  • ( String ) node['cq']['author']['jvm']['debug_opts'] - JVM parameters related to debug interface
  • ( String ) node['cq']['author']['jvm']['crx_opts'] - CRX related JVM parameters
  • ( String ) node['cq']['author']['jvm']['extra_opts'] - All other JVM parameters
  • ( String ) node['cq']['author']['healthcheck']['resource'] - Resource that's queried during instance start to determine whether CQ/AEM is fully operational
  • ( String ) node['cq']['author']['healthcheck']['response_code'] - Expected HTTP status code of healthcheck resource
  • ( String ) node['cq']['author']['healthcheck']['response_body'] - Expected string in HTTP healthcheck response

publish.rb

All attributes in this file refer to CQ/AEM publish instance (node['cq']['publish'] namespace).

  • ( String ) node['cq']['publish']['run_mode'] - Instance run mode
  • ( String ) node['cq']['publish']['port'] - Main port of CQ/AEM instance
  • ( String ) node['cq']['publish']['jmx_ip'] - Value of -Djava.rmi.server.hostname JVM parameter.Requires reference to ${CQ_JMX_IP} shell variable in node['cq']['publish']['jvm']['jmx_opts'] attribute to be effective
  • ( String ) node['cq']['publish']['jmx_port'] - Value of -Dcom.sun.management.jmxremote.port and -Dcom.sun.management.jmxremote.rmi.port JVM parameters. Requires reference to ${CQ_JMX_PORT} shell variable in node['cq']['publish']['jvm']['jmx_opts'] attribute to be effective
  • ( String ) node['cq']['publish']['debug_ip'] - IP to listen on with debug interface. Requires reference to ${CQ_DEBUG_IP} shell variable in node['cq']['publish']['jvm']['debug_opts'] attribute to be effective
  • ( String ) node['cq']['publish']['debug_port'] - Port of JVM debug interface. Requires reference to ${CQ_DEBUG_PORT} shell variable in node['cq']['publish']['jvm']['debug_opts'] attribute to be effective
  • ( String ) node['cq']['publish']['credentials']['login'] - User that's used to perform actions on your CQ/AEM instance. The most typical scenarios require admin
  • ( String ) node['cq']['publish']['credentials']['password'] - Password of user specified in node['cq']['publish']['credentials']['login']
  • ( String ) node['cq']['publish']['jvm']['min_heap'] - -Xms JVM parameter (in megabytes)
  • ( String ) node['cq']['publish']['jvm']['max_heap'] - -Xmx JVM parameter (in megabytes)
  • ( String ) node['cq']['publish']['jvm']['max_perm_size'] - -XX:MaxPermSize JVM parameter (in megabytes)
  • ( String ) node['cq']['publish']['jvm']['code_cache_size'] - -XX:ReservedCodeCacheSize JVM parameter (in megabytes)
  • ( String ) node['cq']['publish']['jvm']['general_opts'] - Generic JVM parameters
  • ( String ) node['cq']['publish']['jvm']['code_cache_opts'] - JVM parameters related to its code cache
  • ( String ) node['cq']['publish']['jvm']['gc_opts'] - JVM parameters related to garbage collection
  • ( String ) node['cq']['publish']['jvm']['jmx_opts'] - JVM parameters related to JMX settings
  • ( String ) node['cq']['publish']['jvm']['debug_opts'] - JVM parameters related to debug interface
  • ( String ) node['cq']['publish']['jvm']['crx_opts'] - CRX related JVM parameters
  • ( String ) node['cq']['publish']['jvm']['extra_opts'] - All other JVM parameters
  • ( String ) node['cq']['publish']['healthcheck']['resource'] - Resource that's queried during instance start to determine whether CQ/AEM is fully operational
  • ( String ) node['cq']['publish']['healthcheck']['response_code'] - Expected HTTP status code of healthcheck resource
  • ( String ) node['cq']['publish']['healthcheck']['response_body'] - Expected string in HTTP healthcheck response

Recipes

default.rb

Installs core dependencies (Ruby gems and OS packages).

commons.rb

Takes care of common elements of every CQ/AEM deployment, including:

  • system user and its configuration
  • required directory structure
  • Java installation
  • CQ Unix Toolkit installation

author.rb

Installs CQ/AEM author instance.

publish.rb

Installs CQ/AEM publish instance.

Custom resources

All CQ/AEM related resource are idempotent, so action won't be taken if not required.

Whenever you need to deploy 2 or more CQ/AEM instances on a single server please make sure you named all your custom resources differently, as you may get unexpected results otherwise (i.e. when CQ/AEM restart is required afterwards). Please find cq_package example below:

Bad:

cq_package 'package1' do
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :deploy
end

cq_package 'package1' do
  instance "http://localhost:#{node['cq']['publish']['port']}"

  action :deploy
end

Good:

cq_package 'Author: package1' do
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :deploy
end

cq_package 'Publish: package1' do
  instance "http://localhost:#{node['cq']['publish']['port']}"

  action :deploy
end

cq_package

Allows for CRX package manipulation using CRX Package Manager API.

Key features:

  • package specific details (name, group, version) are always extracted from ZIP file (/META-INF/vault/properties.xml), so you don't have to define that anywhere else. All you need is an URL to your package
  • cq_package identifies packages by name/group/version properties
  • packages are automatically downloaded from remote (http://, https://) or local (file://) sources. If HTTP(S) source requires basic auth please use http_user and http_pass
  • by default all packages are downloaded to Chef's cache (/var/chef/cache)
  • installation process is considered finished only when both "foreground" (Package Manager) and "background" (OSGi bundle/component restarts) ones are over - no more 'wait until you see X in error.log'

Actions

If you'd like to upload and install a package, in most cases please use deploy action instead of combined upload and install. Detailed explanation can be found below.

  • upload - uploads package to given CQ instance
  • install - installs already uploaded package
  • deploy - uploads and installs given package as a single action. This action is quicker than separate upload + install as less healthchecks have to be executed
  • uninstall - uninstalls given CQ package
  • delete - deletes given CQ package

Properties

  • ( String ) name - Package name. Can be anything as long as it means something to you. Actual package name is extracted from provided ZIP file. Whenever you use notifies on your package resource and more than a single action as defined (i.e. action [:upload, :install]), two notifications will be triggered (after :upload and :install respectively)
  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL
  • ( String ) source - URL to ZIP package. Accepted protocols: file://, http://, https://
  • ( String ) http_user - HTTP basic auth user. Use whenever source requires such authentication
  • ( String ) http_pass - HTTP basic auth password. Use whenever source requires such authentication
  • ( Boolean ) recursive_install - Whether to use recursive flag when installing packages (required for service packs and some hotfixes). Applies only to install and deploy actions
  • ( Boolean ) rescue_mode - Some packages may cause shutdown of the entire OSGi because of dependency (i.e. cycle) or bundle priority issues. In such case after package installation java process is still running, however the instance is not responding over HTTP. After CQ/AEM restart everything works perfectly fine again. This flag allows Chef to continue processing if it is not able to get OSGi bundles state error_state_barrier times in a row. In most (if not all) cases it should be combined with restart notification (please see examples below). It is highly discouraged to use this property, as 99% of CRX packages shouldn't require such configuration. Unfortunately that 1% does. This is rather a safety switch than a common pattern that should be used in every single case. Applies only to install and deploy actions.
  • ( String ) checksum - ZIP file checksum (passed through to remote_file resource that is used under the hood by cq_package provider)
  • ( Integer ) same_state_barrier - How many times in a row the same OSGi state should occur after package (un)installation to consider this process successful. Default is 6
  • ( Integer ) error_state_barrier - How many times in a row the OSGi console was unavailable after package (un)installation. Useful only when combined with rescue_mode. By default set to 6
  • ( Integer ) max_attempts - Number of attempts while waiting for stable OSGi state after package (un)installation. Set to 30 by default
  • ( Integer ) sleep_time - Sleep time between OSGi status checks (in seconds) after package (un)installation. Set to 10 by default

Usage

cq_package 'Slice 4.2.1' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source 'https://oss.sonatype.org/content/groups/public/com/cognifide/slice'\
    '/slice-assembly/4.2.1/slice-assembly-4.2.1-cq.zip'

  action :upload
end

First cq_package resource will download Slice package from provided URL and upload it to defined AEM Author instance.

cq_package 'Upgrade to Oak 1.0.13' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source 'https://artifacts.example.com/aem/6.0.0/hotfixes'\
    '/cq-6.0.0-hotfix-6316-1.1.zip'
  http_user 'john'
  http_pass 'passw0rd'

  action :upload
end

Second resource does the same as the first one, but for Oak 1.0.13 hotfix. The only difference is that provided URL requires basic auth, hence the http_user and http_pass properties.

cq_package 'ACS AEM Commons 1.10.2' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source 'https://github.com/Adobe-Consulting-Services/acs-aem-commons'\
    '/releases/download/acs-aem-commons-1.10.2'\
    '/acs-aem-commons-content-1.10.2.zip'

  action [:upload, :install]
end

Third package shows how to combine multiple actions in a single cq_package resource usage.

cq_package 'AEM6 hotfix 6316' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['aem6']['hf6316']
  recursive_install true

  action :deploy

  notifies :restart, 'service[cq60-author]', :immediately
end

4th cq_package presents how to use deploy action that combines both upload and install in a single execution. This is preferred way of doing package deployment, in particular for those that require AEM service restart as soon as installation is completed. recursive_install was also used here, which is required for majority of hotfixes and every service pack.

cq_package 'Geometrixx All' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source "http://localhost:#{node['cq']['author']['port']}/etc/packages"\
    '/day/cq60/product/cq-geometrixx-all-pkg-5.7.476.zip'
  http_user node['cq']['author']['credentials']['login']
  http_pass node['cq']['author']['credentials']['password']

  action :uninstall
end

Next example describes usage of uninstall action. In this particular case operation was executed against Geometrixx package.

cq_package 'Not really well-thought-out package' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['myapp']
  http_user node['cq']['author']['credentials']['login']
  http_pass node['cq']['author']['credentials']['password']
  rescue_mode true

  action :deploy

  notifies :restart, 'service[cq60-author]', :immediately
end

6th cq_package presents usage of rescue_mode property. Imagine that this package provides new OSGi bundles and right after its installation some serious issue occurs (i.e. unresolvable OSGi dependency, conflict or cycle). As a result of this event all bundles will be turned off and effectively instance will stop responding or start serving 404s for all resources (including /system/console). The java process though will still be running. The only solution to that problem is AEM restart, after which all work perfectly fine again. Without rescue_mode property cq_package provider will keep checking OSGi bundles to detect their stable state, but none of these attempts will end successfully, as nothing is reachable over HTTP. Eventually Chef run will be aborted. If rescue_mode was activated (set to true) then after error_state_barrier unsuccessful attempts an error will be printed and the processing will be continued (restart of cq60-author service in this case).

cq_package 'Author: Service Pack 2 (upload)' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['aem6']['sp2']

  action :upload
end

cq_package 'Author: Service Pack 2 (install)' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['aem6']['sp2']
  recursive_install true
  rescue_mode true
  same_state_barrier 12
  error_state_barrier 12
  max_attempts 36

  action :install

  notifies :restart, 'service[cq60-author]', :immediately
end

7th & 8th cq_package resources explain how to deal with AEM instance restarts after package installation and tune post installation OSGi stability checks.

Moreover it explains how to use combination of upload and install instead of deploy. Such procedure might be required sometimes, i.e. when some extra steps have to be done after package upload, but before its installation.

Please notice that both resources were named differently on purpose to avoid resource merge and 2 restarts. If you'd use:

cq_package 'Author: Service Pack 2' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['aem6']['sp2']
  recursive_install true

  action [:upload, :install]

  notifies :restart, 'service[cq60-author]', :immediately
end

or

cq_package 'Author: Service Pack 2' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['aem6']['sp2']

  action :upload
end

cq_package 'Author: Service Pack 2' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source node['cq']['packages']['aem6']['sp2']
  recursive_install true

  action :install

  notifies :restart, 'service[cq60-author]', :immediately
end

two restarts will be triggered.

In the first case during compile phase Chef will generate 2 resources with the same name, but different actions.

In second example restart still will be triggered after upload, even if it's not explicitly defined during 1st usage (upload action). The reason is quite simple - both resources are named the same (Author: Service Pack 2) and Chef will treat this as a single resource on resource collection. It means that notify parameter will be silently merged to the resource with upload action during compile phase.

cq_osgi_config

Provides an interface for CRUD operations in OSGi configs.

Actions

For regular (non-factory, single instance) configs:

  • create - updates already existing configuration
  • delete - restores default settings of given OSGi config

For factory configs:

  • create - creates a new factory instance if none of existing ones match to defined state
  • delete - deletes factory config instance if there's one that matches to defined state

Properties

  • ( String ) pid - Config name (PID). Relevant to regular configs only
  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL
  • ( String ) factory_pid - Factory PID
  • ( Hash ) properties - Key-value pairs that represent OSGi config properties
  • ( Boolean ) append - If set to true arrays will be merged. Use if you'd like to specify just a subset of array elements. false by default. Has no impact on other property types (String, Integer, etc)
  • ( Boolean ) apply_all - If true all properties defined in a cq_osgi_config resource will be used when applying OSGi configuration (despite of the fact just a subset differs). Example: 5 properties were defined as properties, 3 of them require update, but all of them will be set. false by default
  • ( Boolean ) include_missing - Properties that were NOT defined by user, but exist in OSGi will be included as a part of an update if this property is set to true for regular OSGi configs. For factory configs it behaves almost the same. If new instance needs to be created then defaults defined in factory PID will be used. In case of existing instance update, all missing properties will be based on properties defined in that instance. This is recommended property when you'd like to edit pre-existing factory or regular configs. true by default
  • ( Array ) unique_fields - Property names/keys that define uniqueness of given config. Applicable to factory configs only. By default, all available property keys will be used (defined by factory config on AEM instance). User doesn't need to define that at all, unless you want to cherry pick particular config. It's generally recommended to specify this for every factory OSGi config. Example: log.name key needs to stay unique for your config
  • ( Integer ) count - Number of duplicated instances of given OSGi configuration. 1 by default. Applicable to factory configs only. Useful when duplicated instances are allowed, i.e. each instance specify some sort of a worker and every single one of them has exactly the same set of properties
  • ( Boolean ) enforce_count - Reduces number of duplicated configs if more than count has been found. Applicable to factory configs only. false by default
  • ( Boolean ) force - If true, defined OSGi config is deleted/updated regardless of current settings. Applies to regular OSGi configs only. This violates idempotence, so please use only_if or not_if blocks to prevent constant execution
  • ( Boolean ) rescue_mode - Some config operations may cause shutdown of the entire OSGi because of dependency (i.e. cycle) or bundle/component priority issues. In such case after config update java process is still running, however the instance is not responding over HTTP. After CQ/AEM restart everything works perfectly fine again. This flag allows Chef to continue processing if it is not able to get OSGi component state error_state_barrier times in a row. In most (if not all) cases it should be combined with AEM restart notification. It is highly discouraged to use this property, as 99% of OSGi configs shouldn't require such configuration. Unfortunately that 1% does. This is rather a safety switch than a common pattern that should be used in every single case.
  • ( Integer ) same_state_barrier - How many times in a row the same OSGi component state should occur after configuration update to consider this process successful. 3 by default
  • ( Integer ) error_state_barrier - How many times in a row the OSGi console was unavailable after OSGi config update. Useful only when combined with rescue_mode. 3 by default
  • ( Integer ) max_attempts - Number of attempts while waiting for stable OSGi state after OSGi config update. 60 by default
  • ( Integer ) sleep_time - Sleep time between OSGi component status checks (in seconds) after config update. 2 by default

Compatibility matrix

Property Regular OSGi config Factory OSGi config
pid
username
password
instance
factory_pid
properties
append
apply_all
include_missing
unique_fields
count
enforce_count
force

Usage

Regular OSGi configs

cq_osgi_config 'Root Mapping' do
  pid 'com.day.cq.commons.servlets.RootMappingServlet'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  properties('rootmapping.target' => '/welcome.html')

  action :create
end

Root Mapping resource sets / redirect to /welcome.html if it's not already set.

cq_osgi_config 'Event Admin' do
  pid 'org.apache.felix.eventadmin.impl.EventAdmin'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  append true
  properties(
    'org.apache.felix.eventadmin.IgnoreTimeout' => ['com.example*']
  )

  action :create
end

Event Admin merges defined properties with the ones that are already set (because of append property). This is how Event Admin will look like before:

ID VALUE
org.apache.felix.eventadmin.ThreadPoolSize 20
org.apache.felix.eventadmin.Timeout 5000
org.apache.felix.eventadmin.RequireTopic true
org.apache.felix.eventadmin.IgnoreTimeout ["org.apache.felix*","com.adobe*"]

and after Chef run:

ID VALUE
org.apache.felix.eventadmin.ThreadPoolSize 20
org.apache.felix.eventadmin.Timeout 5000
org.apache.felix.eventadmin.RequireTopic true
org.apache.felix.eventadmin.IgnoreTimeout ["com.adobe*","com.example*","org.apache.felix*"]
cq_osgi_config 'OAuth Twitter' do
  pid 'com.adobe.granite.auth.oauth.impl.TwitterProviderImpl'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  properties(
    'oauth.provider.id' => 'twitter'
  )

  action :delete
end

Properties of OAuth Twitter will be restored to default values if any of them was previously modified (explicitly set).

cq_osgi_config 'Promotion Manager' do
  pid 'com.adobe.cq.commerce.impl.promotion.PromotionManagerImpl'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  force true

  action :delete
end

Promotion Manager will behave as OAuth Twitter with one exception - it will happen on every Chef run due to force property.

Factory OSGi configs

cq_osgi_config 'Custom Logger' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  factory_pid 'org.apache.sling.commons.log.LogManager.factory.config'
  properties(
    'org.apache.sling.commons.log.level' => 'error',
    'org.apache.sling.commons.log.file' => 'logs/custom.log',
    'org.apache.sling.commons.log.pattern' =>
      '{0,date,dd.MM.yyyy HH:mm:ss.SSS} *{4}* [{2}] {3} {5}',
    'org.apache.sling.commons.log.names' => [
      'com.example.custom1',
      'com.example.custom2'
    ]
  )
  unique_fields ['org.apache.sling.commons.log.file']

  action :create
end

Custom Logger resource will create a new logger according to defined properties. org.apache.sling.commons.log.file is a virtual identifier of given OSGi config instance (specified by user). If instance with such "ID" already exists nothing happens. Otherwise a brand new configuration will be created. Please keep in mind that there's no need to specify any UUID in resource definition.

cq_osgi_config 'Job Queue' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  factory_pid 'org.apache.sling.event.jobs.QueueConfiguration'
  properties(
    'queue.name' => 'Granite Workflow Timeout Queue',
    'queue.type' => 'TOPIC_ROUND_ROBIN',
    'queue.topics' => ['com/adobe/granite/workflow/timeout/job'],
    'queue.maxparallel' => -1,
    'queue.retries' => 10,
    'queue.retrydelay' => 2000,
    'queue.priority' => 'MIN',
    'service.ranking' => 0
  )
  unique_fields ['queue.name']

  action :delete
end

Job Queue resource will delete factory instance of org.apache.sling.event.jobs.QueueConfiguration that has queue.name set to Granite Workflow Timeout Queue. Presence of additional properties doesn't matter in this case and will be completely ignored.

cq_osgi_bundle

Adds ability to stop and start OSGi bundles

Actions

  • stop - stop given OSGi bundle if it is in Active state
  • start - starts defined bundle, but only when it's in Resolved state

Properties

  • ( String ) symbolic_name - Symbolic name of the bundle, i.e. com.company.example.abc. If not explicitly defined resource name will be used as symbolic name
  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL
  • ( Boolean ) rescue_mode - Same meaning as for cq_package
  • ( Integer ) same_state_barrier - Same meaning as for cq_package
  • ( Integer ) error_state_barrier - Same meaning as for cq_package
  • ( Integer ) max_attempts - Same meaning as for cq_package
  • ( Integer ) sleep_time - Same meaning as for cq_package

Usage

cq_osgi_bundle 'Author: org.eclipse.equinox.region' do
  symbolic_name 'org.eclipse.equinox.region'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  same_state_barrier 3
  sleep_time 5

  action :stop
end

First example stops org.eclipse.equinox.region AEM author instance. Since there's just a few dependencies on this bundle, number of post-stop checks have been limited, as there's no point to wait for so long.

cq_osgi_bundle 'com.adobe.xmp.worker.files.native.fragment.linux' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :start
end

Second instance of cq_osgi_bundle is fairly simple, as it just starts com.adobe.xmp.worker.files.native.fragment.linux bundle.

cq_osgi_component

Please keep in mind that OSGi components used to get back to their original state after AEM instance restart. If you disabled one, most probably it'll become enabled after instance restart.

Actions

  • enable - enable given OSGi component
  • disable - disable defined OSGi component

Properties

  • ( String ) pid - Component PID
  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL

Usage

cq_osgi_component 'Author: com.example.my.component' do
  pid 'com.example.my.component'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :enable
end

cq_osgi_component 'Author: com.project.email.servlet' do
  pid 'com.project.email.servlet'
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :disable
end

Both examples are self-explanatory. First one enables com.example.my.component component if it's in disabled state. Second one will disable com.project.email.servlet component, but only if it's state is not already disabled.

cq_user

Exposes a resource for CQ/AEM user management. Supports:

  • password updates
  • profile updates (e-mail, job title, etc)
  • status updates (activate/deactivate given user)

Actions

  • modify - use to modify an existing user. Action will be skipped if given user does not exist

Properties

  • ( String ) id - User ID (login)
  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL
  • ( String ) email - E-mail
  • ( String ) first_name - First name
  • ( String ) last_name - Last name
  • ( String ) phone_number - Phone number
  • ( String ) job_title - Job title
  • ( String ) street - Street
  • ( String ) mobile - Mobile
  • ( String ) city - City
  • ( String ) postal_code - Postal code
  • ( String ) country - Country
  • ( String ) state - State
  • ( String ) gender - Gender
  • ( String ) about - About section
  • ( String ) user_password - Desired password for non-admin user specified by id property
  • ( Boolean ) enabled - true by default, set to false to deactivate given user. Has no effect for admin user
  • ( String ) old_password - Old password of admin user. Has no effect for non-admin ones

Compatibility matrix

Property admin user All other users
id
username
password
instance
email
first_name
last_name
phone_number
job_title
street
mobile
city
postal_code
country
state
gender
about
user_password
enabled
old_password

Usage

cq_user 'admin' do
  username node['cq']['author']['credentials']['login']
  password 'd4rk_kn1ght'
  instance "http://localhost:#{node['cq']['author']['port']}"

  first_name 'Bruce'
  last_name 'Wayne'
  old_password 'passw0rd'

  action :modify
end

Modify action on cq_user 'admin' resource will change CQ/AEM admin password to d4rk_kn1ght if the current one is either passw0rd or admin (the latter is automatically checked if both password and old_password are incorrect). Moreover admin first name and last name will be updated (to Bruce and Wayne respectively) if needed.

cq_user 'author' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"

  first_name 'Peter'
  last_name 'Parker'
  job_title 'Spiderman'
  gender 'male'
  enabled false
  user_password 'sp1d3r'

  action :modify
end

Second example (cq_user 'author') also updates user password, but this time the old one doesn't have to be specified, as this operation will be executed on admin rights (auth credentials: username/password). Additionally auhtor's profile will be updated and user will be disabled (enabled false), so you won't be able to log in as this user anymore.

cq_jcr

CRUD operations on JCR nodes.

Actions

  • create - creates new node under given path if it doesn't exist. Otherwise it modifies its properties if required
  • delete - deletes node if it exists. Prints error otherwise
  • modify - modifies properties of existing JCR node

Properties

  • ( String ) path - Node path
  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL
  • ( Hash ) properties - Node properties
  • ( Boolean ) append - By default set to true. If full overwrite of properties is required please set append property to false. Applies only to :create and :modify actions

Usage

cq_jcr '/content/test_node' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  properties(
    'property_one' => 'first',
    'property_two' => 'second',
    'property_three' => ['item1', 'item2', 'item3']
  )

  action :create
end

Create action on cq_jcr '/content/test_node' will create such node with given properties if it doesn't exist yet. Otherwise its properties will be updated if necessary. By default append is set to true, which means existing properties of /content/test_node will stay untouched unless the same properties are specified in your cq_jcr resource.

cq_jcr '/content/geometrixx/en/products/jcr:content' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  append false
  properties(
    'jcr:primaryType' => 'cq:PageContent',
    'jcr:title' => 'New title',
    'subtitle' => 'New subtitle',
    'new_property' => 'Random value'
  )

  action :create
end

2nd example sets append property to false, which means that all properties except those specified in your resource should be removed. It will act as a full overwrite (keep in mind that some properties are protected and can't be deleted, moreover Sling API automatically adds things like jcr:createdBy).

cq_jcr '/content/dam/geometrixx-media/articles/en/2012' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :delete
end

Next example is quite simple - /content/dam/geometrixx-media/articles/en/2012 will get deleted if it exists. Otherwise warning message will be printed.

cq_jcr '/content/geometrixx/en/services/certification/jcr:content' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  properties(
    'jcr:title' => 'New Certification Services',
    'brand_new_prop' => 'ValueX'
  )

  action :modify
end

Last cq_jcr resource uses :modify action. It applies updates to existing nodes only. If specified path does not exist warning message will be displayed.

cq_start_guard

Allows you to wait for full AEM instance start before moving on with subsequent operations. It periodically sends HTTP request to AEM and compares response (both status code and body) with expected state. As soon as defined requirements are met the resource stops its job.

Actions

  • nothing - default action, does nothing :)
  • run - verifies instance state according to defined properties. This action is NOT idempotent by design and should be always triggered via notify from other resources

Properties

  • ( String ) name - Start guard name
  • ( String ) instance - Instance URL
  • ( String ) path - URL path that's requested to verify instance health
  • ( String ) expected_code - Expected HTTP status code
  • ( String ) expected_body - Expected string in HTTP response body
  • ( Integer ) timeout - Maximum time in seconds before giving up
  • ( Integer ) http_timeout - Maximum time for HTTP call
  • ( Integer ) interval - Time in seconds between HTTP healthcheck request attempts

Usage

service 'cq64-author' do
  supports status: true, restart: true
  action :start

  notifies :run, "cq_start_guard[cq64-author]", :immediately
end

cq_start_guard 'cq64-author' do
  instance "http://localhost:#{node['cq']['author']['port']}"
  path node['cq']['author']['healthcheck']['resource']
  expected_code node['cq']['author']['healthcheck']['response_code']
  expected_body node['cq']['author']['healthcheck']['response_body']
  timeout node['cq']['service']['start_timeout']

  action :nothing
end

Whenever AEM gets (re)started run cq_start_guard and wait until /libs/granite/core/content/login.html returns 200 response code

service 'cq64-author' do
  supports status: true, restart: true
  action :start

  notifies :run, "cq_start_guard[cq64-author]", :immediately
end

cq_start_guard 'cq64-author' do
  instance "http://localhost:#{node['cq']['author']['port']}"
  path '/bin/healthchecks/instance'
  expected_code '200'
  expected_body '{"status": "ok"}'
  timeout 900
  http_timeout 5
  interval 10

  action :nothing
end

Right after restart of cq64-author service send notification to cq_start_guard and wait until /bin/healthchecks/instance returns 200 code and {"status": "ok"} JSON in the body. Don't spend more than 15 minutes on such health check. Requests will be send every 10 seconds, however each HTTP call can't last more than 5 seconds.

cq_clientlib_cache

This resource enables invalidation/rebuilt of internal clientlib cache in AEM. Please keep in mind that cq_clientlib_cache is not idempotent and it is generally recommended to trigger it via notify from other resources.

Actions

  • nothing - default action
  • invalidate - invalidates the entire clientlib cache
  • rebuild - rebuilds all clientlibs (please keep in mind this operation usually takes at least a couple of minutes)

Properties

  • ( String ) username - Instance username
  • ( String ) password - Instance password
  • ( String ) instance - Instance URL

Usage

cq_package 'Custom AEM app' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"
  source 'http://artifacts.example.org/app/1.0/myapp-1.0.zip'
  recursive_install true

  action :deploy

  notifies :invalidate, 'cq_clientlib_cache[invalidation]', :delayed
end

cq_clientlib_cache 'invalidation' do
  username node['cq']['author']['credentials']['login']
  password node['cq']['author']['credentials']['password']
  instance "http://localhost:#{node['cq']['author']['port']}"

  action :nothing
end

Testing

TBD

Author

Jakub Wadolowski (jakub.wadolowski@cognifide.com)