This cookbook is used to configure a system as a Chef Client.
- AIX 6+
- Clear Linux
- Debian
- Fedora
- FreeBSD
- Mac OS X
- openSUSE
- SLES 12+
- RHEL
- Solaris 10+
- Ubuntu
- Windows 2008 R2+
- Chef 13.0+
- cron 2.0+
- logrotate 1.9.0+
See USAGE.
The following attributes affect the behavior of the chef-client program when running as a service through one of the service recipes, or in cron with the cron recipe, or are used in the recipes for various settings that require flexibility.
node['chef_client']['interval']
- SetsChef::Config[:interval]
via command-line option for number of seconds between chef-client daemon runs. Default 1800.node['chef_client']['splay']
- SetsChef::Config[:splay]
via command-line option for a random amount of seconds to add to interval. Default 300.node['chef_client']['log_file']
- Sets the file name used to store chef-client logs. Default "client.log".node['chef_client']['log_dir']
- Sets directory used to store chef-client logs. Default "/var/log/chef".node['chef_client']['log_rotation']['options']
- Set options to logrotation of chef-client log file. Default['compress']
.node['chef_client']['log_rotation']['prerotate']
- Set prerotate action for chef-client logrotation. Default tonil
.node['chef_client']['log_rotation']['postrotate']
- Set postrotate action for chef-client logrotation. Default to chef-client service reload depending on init system.node['chef_client']['conf_dir']
- Sets directory used via command-line option to a location where chef-client search for the client config file . Default "/etc/chef".node['chef_client']['bin']
- Sets the full path to thechef-client
binary. Mainly used to set a specific path if multiple versions of chef-client exist on a system or the bin has been installed in a non-sane path. Default "/usr/bin/chef-client".node['chef_client']['ca_cert_path']
- Sets the full path to the PEM-encoded certificate trust store used bychef-client
when daemonized. If not set, default values are used.node['chef_client']['cron']['minute']
- The minute that chef-client will run as a cron task. See cron recipenode['chef_client']['cron']['hour']
- The hour that chef-client will run as a cron task. See cron recipenode['chef_client']['cron']['weekday']
- The weekday that chef-client will run as a cron task. See cron recipenode['chef_client']['cron']['environment_variables']
- Environment variables to pass to chef-client's execution (e.g.SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt
chef-client)node['chef_client']['cron']['log_file']
- Location to capture the log output of chef-client during the chef run.node['chef_client']['cron']['append_log']
- Whether to append to the log. Default:false
chef-client output.node['chef_client']['cron']['use_cron_d']
- If true, use thecron_d
resource. If false (default), use the cron resource built-in to Chef.node['chef_client']['cron']['mailto']
- If set,MAILTO
env variable is set for cron definitionnode['chef_client']['cron']['priority']
- If set, defines the scheduling priority for thechef-client
process. MUST be a value between -20 and 19. ONLY applies to *nix-style operating systems.node['chef_client']['reload_config']
- If true, reload Chef config of current Chef run whenclient.rb
template changes (defaults to true)node['chef_client']['daemon_options']
- An array of additional options to pass to the chef-client service, empty by default, and must be an array if specified.node['chef_client']['systemd']['timer']
- If true, uses systemd timer to run chef frequently instead of chef-client daemon mode (defaults to false). This only works on platforms where systemd is installed and used.node['chef_client']['systemd']['timeout']
- If configured, sets the systemd timeout. This might be useful to avoid stalled chef runs in the systemd timer setup.node['chef_client']['systemd']['restart']
- The string to use for systemdRestart=
value when not running as a timer. Defaults toalways
. Other possible options:no, on-success, on-failure, on-abnormal, on-watchdog, on-abort
.node['chef_client']['task']['frequency']
- Frequency with which to run thechef-client
scheduled task (e.g.,'hourly'
,'daily'
, etc.) Default is'minute'
.node['chef_client']['task']['frequency_modifier']
- Numeric value to go with the scheduled task frequency. Default isnode['chef_client']['interval'].to_i / 60
node['chef_client']['task']['start_time']
- The start time for the task inHH:mm
format (ex: 14:00). If thefrequency
isminute
default start time will beTime.now
plus thefrequency_modifier
number of minutes.node['chef_client']['task']['start_date']
- The start date for the task inm:d:Y
format (ex: 12/17/2017). nil by default and isn't necessary if you're running a regular interval.node['chef_client']['task']['user']
- The user the scheduled task will run as, defaults to'SYSTEM'
.node['chef_client']['task']['password']
- The password for the user the scheduled task will run as, defaults tonil
because the default user,'SYSTEM'
, does not need a password.node['chef_client']['task']['name']
- The name of the scheduled task, defaults tochef-client
.
The following attributes are set on a per-platform basis, see the attributes/default.rb
file for default values.
-
node['chef_client']['init_style']
- Sets up the client service based on the style of init system to use. Default is based on platform and falls back to'none'
. See service recipes. -
node['chef_client']['run_path']
- Directory location where chef-client should write the PID file. Default based on platform, falls back to "/var/run". -
node['chef_client']['cache_path']
- Directory location for -
Chef::Config[:file_cache_path]
where chef-client will cache various files. Default is based on platform, falls back to "/var/chef/cache". -
node['chef_client']['backup_path']
- Directory location forChef::Config[:file_backup_path]
where chef-client will backup templates and cookbook files. Default is based on platform, falls back to "/var/chef/backup". -
node['chef_client']['launchd_mode']
- (Only for Mac OS X) if set to'daemon'
, runs chef-client with-d
and-s
options; defaults to'interval'
. -
When
chef_client['log_file']
is set and running on a logrotate supported platform (debian, rhel, fedora family), use the following attributes to tune log rotation.node['chef_client']['logrotate']['rotate']
- Number of rotated logs to keep on disk, default 12.node['chef_client']['logrotate']['frequency']
- How often to rotate chef client logs, default weekly.
This cookbook makes use of attribute-driven configuration with this attribute. See USAGE for examples.
node['chef_client']['config']
- A hash of Chef::Config keys and their values, rendered dynamically in/etc/chef/client.rb
.node['chef_client']['load_gems']
- Hash of gems to load into chef via the client.rb filenode['ohai']['disabled_plugins']
- An array of ohai plugins to disable, empty by default, and must be an array if specified. Ohai 6 plugins should be specified as a string (ie. "dmi"). Ohai 7+ plugins should be specified as a symbol within quotation marks (ie. ":Passwd").node['ohai']['plugin_path']
- An additional path to load Ohai plugins from. Necessary if you use the ohai_plugin resource in the Ohai cookbook to install your own ohai plugins.
For the most current information about Chef Client configuration, read the documentation..
-
node['chef_client']['config']['chef_server_url']
- The URL for the Chef server. -
node['chef_client']['config']['validation_client_name']
- The name of the chef-validator key that is used by the chef-client to access the Chef server during the initial chef-client run. -
node['chef_client']['config']['verbose_logging']
- Set the log level. Options: true, nil, and false. When this is set to false, notifications about individual resources being processed are suppressed (and are output at the :info logging level). Setting this to false can be useful when a chef-client is run as a daemon. Default value: nil. -
node['chef_client']['config']['rubygems_url']
- The location to source rubygems. It can be set to a string or array of strings for URIs to set as rubygems sources. This allows individuals to setup an internal mirror of rubygems for "airgapped" environments. Default value:https://www.rubygems.org
. -
See USAGE for how to set handlers with the
config
attribute.
This section describes the recipes in the cookbook and how to use them in your environment.
Sets up the /etc/chef/client.rb
config file from a template and reloads the configuration for the current chef-client run.
See USAGE for more information on how the configuration is rendered with attributes.
The chef-client::service
recipe includes one of the chef-client::INIT_STYLE_service
recipes based on the attribute, node['chef_client']['init_style']
. The individual service recipes can be included directly, too. For example, to use the init scripts, on a node or role's run list:
recipe[chef-client::init_service]
Use this recipe on systems that should have a chef-client
daemon running, such as when Knife bootstrap was used to install Chef on a new system.
init
- uses the init script included in this cookbook, supported on debian and redhat family distributions.upstart
- uses the upstart job included in this cookbook, supported on ubuntu.launchd
- sets up the service under launchd, supported on Mac OS X & Mac OS X Server.bsd
- prints a message about how to update BSD systems to enable the chef-client service.systemd
- sets up the service under systemd. Supported on systemd based distros.
Includes the chef-client::service
recipe by default on *nix platforms and the task recipe for Windows hosts.
Use this recipe to delete the validation certificate (default /etc/chef/validation.pem
) when using a chef-client
after the client has been validated and authorized to connect to the server.
Use this recipe to run chef-client as a cron job rather than as a service. The cron job runs after random delay that is between 0 and 90 seconds to ensure that the chef-clients don't attempt to connect to the chef-server at the exact same time. You should set node['chef_client']['init_style'] = 'none'
when you use this mode but it is not required.
Use this recipe to run chef-client on Windows nodes as a scheduled task. Without modifying attributes the scheduled task will run 30 minutes after the recipe runs, with each chef run rescheduling the run 30 minutes in the future. By default the job runs as the system user. The time period between runs can be modified with the default['chef_client']['task']['frequency_modifier']
attribute and the user can be changed with the default['chef_client']['task']['user']
and default['chef_client']['task']['password']
attributes.
Use the recipes as described above to configure your systems to run Chef as a service via cron / scheduled task or one of the service management systems supported by the recipes.
The chef-client::config
recipe is only required with init style init
(default setting for the attribute on debian/redhat family platforms, because the init script doesn't include the pid_file
option which is set in the config.
The config recipe is used to dynamically generate the /etc/chef/client.rb
config file. The template walks all attributes in node['chef_client']['config']
and writes them out as key:value pairs. The key should be the configuration directive. For example, the following attributes (in a role):
default_attributes(
"chef_client" => {
"config" => {
"ssl_verify_mode" => ":verify_peer",
"client_fork" => true
}
}
)
will render the following configuration (/etc/chef/client.rb
):
chef_server_url "https://api.chef.io/organizations/MYORG"
validation_client_name "MYORG-validator"
ssl_verify_mode :verify_peer
node_name "config-ubuntu-1204"
client_fork true
The chef_server_url
, node_name
and validation_client_name
are set by default in the attributes file from Chef::Config
. They are presumed to come from the knife bootstrap
command when setting up a new node for Chef. To set the node name to the default value (the node['fqdn']
attribute), it can be set false. Be careful when setting this or the Server URL, as those values may already exist.
As another example, to set HTTP proxy configuration settings. By default Chef will not use a proxy.
default_attributes(
"chef_client" => {
"config" => {
"http_proxy" => "http://proxy.mycorp.com:3128",
"https_proxy" => "http://proxy.mycorp.com:3128",
"http_proxy_user" => "my_username",
"http_proxy_pass" => "Awe_some_Pass_Word!",
"no_proxy" => "*.vmware.com,10.*"
}
}
)
The /etc/chef/client.rb
file will include all the configuration files in /etc/chef/client.d/*.rb
. To create custom configuration, simply render a file resource with file
(and the content
parameter), template
, remote_file
, or cookbook_file
. For example, in your own cookbook that requires custom Chef client configuration, create the following cookbook_file
resource:
chef_gem 'syslog-logger'
cookbook_file "/etc/chef/client.d/myconfig.rb" do
source "myconfig.rb"
mode '0644'
notifies :create, "ruby_block[reload_client_config]"
end
include_recipe 'chef-client::config'
Then create files/default/myconfig.rb
with the configuration content you want. For example, if you wish to create a configuration to log to syslog:
require 'syslog-logger'
require 'syslog'
Logger::Syslog.class_eval do
attr_accessor :sync, :formatter
end
log_location Chef::Log::Syslog.new('chef-client', ::Syslog::LOG_DAEMON)
On Windows:
log_location Chef::Log::WinEvt.new
Use the load_gems
attribute to install gems that need to be required in the client.rb. This attribute should be a hash. The gem will also be installed with chef_gem
. For example, suppose we want to use a Chef Handler Gem, chef-handler-updated-resources
, which is used in the next heading. Set the attributes, e.g., in a role:
default_attributes(
"chef_client" => {
"load_gems" => {
"chef-handler-updated-resources" => {
"require_name" => "chef/handler/updated_resources",
"version" => "0.1"
}
}
}
)
Each key in load_gems
is the name of a gem. Each gem hash can have two keys, the require_name
which is the string that will be require
'd in /etc/chef/client.rb
, and version
which is the version of the gem to install. If the version is not specified, the latest version will be installed.
The above example will render the following in /etc/chef/client.rb
:
["chef/handler/updated_resources"].each do |lib|
begin
require lib
rescue LoadError
Chef::Log.warn "Failed to load #{lib}. This should be resolved after a chef run."
end
end
To dynamically render configuration for Start, Report, or Exception handlers, set the following attributes in the config
attributes:
start_handlers
report_handlers
exception_handlers
This is an alternative to using the chef_handler
cookbook.
Each of these attributes must be an array of hashes. The hash has two keys, class
(a string), and arguments
(an array). For example, to use the report handler in the Requiring Gems section:
default_attributes(
"chef_client" => {
"config" => {
"report_handlers" => [
{"class" => "SimpleReport::UpdatedResources", "arguments" => []}
]
}
}
)
If the handler you're using has an initialize method that takes arguments, then pass each one as a member of the array. Otherwise, leave it blank as above.
This will render the following in /etc/chef/client.rb
.
report_handlers << SimpleReport::UpdatedResources.new()
On Mac OS X and Mac OS X Server, the default service implementation is "launchd".
Since launchd can run a service in interval mode, by default chef-client is not started in daemon mode like on Debian or Ubuntu. Keep this in mind when you look at your process list and check for a running chef process! If you wish to run chef-client in daemon mode, set attribute chef_client.launchd_mode
to "daemon".
This cookbook does not handle updating the chef-client, as that's out of the cookbook's current scope. To sensibly manage updates of the chef-client's install, we refer you to:
- chef_client_updater - Cookbook for keeping your install up-to-date
The chef_client_scheduled_task setups up chef-client to run as a scheduled task. This resource is what the task recipe calls under the hood. You can use this recipe directly when writing a wrapper cookbook. Additionally using this resource directly allows you to control where you store the user credentials instead of storing them as node attributes. This is useful if you want to store these credentials in an encrypted databag.
:add
:remove
user
- The username to run the task as. default: 'System'password
, The password of the user to run the task as if not using the System userfrequency
- Frequency with which to run the task (e.g., 'hourly', 'daily', etc.) Default is 'minute'frequency_modifier
Numeric value to go with the scheduled task frequency - default: '30'start_time
The start time for the task in HH:mm format (ex: 14:00). If thefrequency
isminute
default start time will beTime.now
plus thefrequency_modifier
number of minutes.start_date
- The start date for the task inm:d:Y
format (ex: 12/17/2017). nil by default and isn't necessary if you're running a regular interval.splay
- A random number of seconds between 0 and X to add to interval. default: '300'config_directory
- The path to the Chef config directory. default: 'C:/chef'log_directory
- The path to the Chef log directory. default: 'CONFIG_DIRECTORY/log'chef_binary_path
- The path to the chef-client binary. default: 'C:/opscode/chef/bin/chef-client'daemon_options
- An optional array of extra options to pass to the chef-clienttask_name
- The name of the scheduled task. This allows for multiple chef_client_scheduled_task resources when it is used directly like in a wrapper cookbook. default: 'chef-client'
This cookbook is maintained by Chef's Community Cookbook Engineering team. Our goal is to improve cookbook quality and to aid the community in contributing to cookbooks. To learn more about our team, process, and design goals see our team documentation. To learn more about contributing to cookbooks like this see our contributing documentation, or if you have general questions about this cookbook come chat with us in #cookbok-engineering on the Chef Community Slack
Copyright: 2010-2017, Chef Software, Inc.
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.