puppet-filebeat

puppet-filebeat - Table of Contents

Description

The filebeat module installs and configures the filebeat log shipper maintained by elastic.

Setup

What filebeat affects

By default filebeat adds a software repository to your system, and installs filebeat along with required configurations.

Upgrading to Filebeat 7.x

To upgrade to Filebeat 7.x, simply set $filebeat::major_version to 7 and $filebeat::package_ensure to latest (or whichever version of 7.x you want, just not present).

You'll also need to change instances of filebeat::prospector to filebeat::input when upgrading to version 4.x of this module.

Setup Requirements

The filebeat module depends on puppetlabs/stdlib, and on puppetlabs/apt on Debian based systems.

Beginning with filebeat

filebeat can be installed with puppet module install pcfens-filebeat (or with r10k, librarian-puppet, etc.)

The only required parameter, other than which files to ship, is the outputs parameter.

Usage

All of the default values in filebeat follow the upstream defaults (at the time of writing).

To ship files to elasticsearch:

class { 'filebeat':
  outputs => {
    'elasticsearch' => {
     'hosts' => [
       'http://localhost:9200',
       'http://anotherserver:9200'
     ],
     'loadbalance' => true,
     'cas'         => [
        '/etc/pki/root/ca.pem',
     ],
    },
  },
}

To ship log files through logstash:

class { 'filebeat':
  outputs => {
    'logstash'     => {
     'hosts' => [
       'localhost:5044',
       'anotherserver:5044'
     ],
     'loadbalance' => true,
    },
  },
}

Shipper and logging options can be configured the same way, and are documented on the elastic website.

Adding an Input

Inputs are processes that ship log files to elasticsearch or logstash. They can be defined as a hash added to the class declaration (also used for automatically creating input using hiera), or as their own defined resources.

At a minimum, the paths parameter must be set to an array of files or blobs that should be shipped. doc_type is what logstash views as the type parameter if you'd like to apply conditional filters.

filebeat::input { 'syslogs':
  paths    => [
    '/var/log/auth.log',
    '/var/log/syslog',
  ],
  doc_type => 'syslog-beat',
}

Multiline Logs

Filebeat inputs can handle multiline log entries. The multiline parameter accepts a hash containing pattern, negate, match, max_lines, and timeout as documented in the filebeat configuration documentation.

JSON Logs

Filebeat inputs (versions >= 5.0) can natively decode JSON objects if they are stored one per line. The json parameter accepts a hash containing message_key, keys_under_root, overwrite_keys, and add_error_key.

Depending on the version, expand_keys, document_id and ignore_decoding_error may be supported as well.

See the filebeat configuration documentation for details.

Inputs in Hiera

Inputs can be defined in hiera using the inputs parameter. By default, hiera will not merge input declarations down the hiera hierarchy. That behavior can be changed by configuring the lookup_options flag.

inputs can be a Hash that will follow all the parameters listed on this documentation or an Array that will output as is to the input config file.

Usage on Windows

When installing on Windows, this module will download the windows version of Filebeat from elastic to C:\Temp by default. The directory can be overridden using the tmp_dir parameter. tmp_dir is not managed by this module, but is expected to exist as a directory that puppet can write to.

Processors

Filebeat 5.0 and greater includes a new libbeat feature for filtering and/or enhancing all exported data through processors before being sent to the configured output(s). They can be defined as a hash added to the class declaration (also used for automatically creating processors using hiera), or as their own defined resources.

To drop the offset and input_type fields from all events:

class {'filebeat':
  processors => [
    {
      'drop_fields' => {
        'fields' => ['input_type', 'offset'],
      }
    }
  ],
}

To drop all events that have the http response code equal to 200: input

class {'filebeat':
  processors => [
    {
      'drop_event' => {
        'when' => {'equals' => {'http.code' => 200}}
      }
    }
  ],
}

Now to combine these examples into a single definition:

class {'filebeat':
  processors => [
    {
      'drop_fields' => {
        'params'   => {'fields' => ['input_type', 'offset']},
        'priority' => 1,
      }
    },
    {
      'drop_event' => {
        'when'     => {'equals' => {'http.code' => 200}},
        'priority' => 2,
      }
    }
  ],
}

For more information please review the documentation here.

Processors in Hiera

Processors can be declared in hiera using the processors parameter. By default, hiera will not merge processor declarations down the hiera hierarchy. That behavior can be changed by configuring the lookup_options flag.

Index Lifecycle Management

You can override the default filebeat ILM policy by specifying ilm.policy hash in filebeat::setup parameter:

filebeat::setup:
  ilm.policy:
    phases:
      hot:
        min_age: "0ms"
        actions:
          rollover:
            max_size: "10gb"
            max_age: "1d"

Reference

puppet-filebeat - Table of Contents

Public Classes

Class: `filebeat`

Installs and configures filebeat.

Parameters within filebeat

package_ensure: [String] The ensure parameter for the filebeat package If set to absent, inputs and processors passed as parameters are ignored and everything managed by puppet will be removed. (default: present)
manage_package: [Boolean] Whether ot not to manage the installation of the package (default: true)
manage_repo: [Boolean] Whether or not the upstream (elastic) repo should be configured or not (default: true)
major_version: [Enum] The major version of Filebeat to install. Should be either '5' or '6'. The default value is '6', except for OpenBSD 6.3 and earlier, which has a default value of '5'.
service_ensure: [String] The ensure parameter on the filebeat service (default: running)
service_enable: [String] The enable parameter on the filebeat service (default: true)
param repo_priority: [Integer] Repository priority. yum and apt supported (default: undef)
service_provider: [String] The provider parameter on the filebeat service (default: on RedHat based systems use redhat, otherwise undefined)
spool_size: [Integer] How large the spool should grow before being flushed to the network (default: 2048)
idle_timeout: [String] How often the spooler should be flushed even if spool size isn't reached (default: 5s)
publish_async: [Boolean] If set to true filebeat will publish while preparing the next batch of lines to transmit (default: false)
config_file: [String] Where the configuration file managed by this module should be placed. If you think you might want to use this, read the limitations first. Defaults to the location that filebeat expects for your operating system.
config_dir: [String] The directory where inputs should be defined (default: /etc/filebeat/conf.d)
config_dir_mode: [String] The permissions mode set on the configuration directory (default: 0755)
config_dir_owner: [String] The owner of the configuration directory (default: root). Linux only.
config_dir_group: [String] The group of the configuration directory (default: root). Linux only.
config_file_mode: [String] The permissions mode set on configuration files (default: 0644)
config_file_owner: [String] The owner of the configuration files, including inputs (default: root). Linux only.
config_file_group: [String] The group of the configuration files, including inputs (default: root). Linux only.
purge_conf_dir: [Boolean] Should files in the input configuration directory not managed by puppet be automatically purged
enable_conf_modules: [Boolean] Should filebeat.config.modules be enabled
modules_dir: [String] The directory where module configurations should be defined (default: /etc/filebeat/modules.d)
cloud: [Hash] Will be converted to YAML for the optional cloud.id and cloud.auth of the configuration (see documentation, and above)
features: [Hash] Will be converted to YAML for the optional features section of the configuration (see documentation, and above)
queue: [Hash] Will be converted to YAML for the optional queue.mem and queue.disk of the configuration (see documentation, and above)
outputs: [Hash] Will be converted to YAML for the required outputs section of the configuration (see documentation, and above)
shipper: [Hash] Will be converted to YAML to create the optional shipper section of the filebeat config (see documentation)
autodiscover: [Hash] Will be converted to YAML for the optional autodiscover section of the configuration (see documentation, and above)`
logging: [Hash] Will be converted to YAML to create the optional logging section of the filebeat config (see documentation)
systemd_beat_log_opts_override: [String] Will overide the default BEAT_LOG_OPTS=-e. Required if using logging hash on systems running with systemd. required: Puppet 6.1+, Filebeat 7+,
modules: [Array] Will be converted to YAML to create the optional modules section of the filebeat config (see documentation)
conf_template: [String] The configuration template to use to generate the main filebeat.yml config file.
download_url: [String] The URL of the zip file that should be downloaded to install filebeat (windows only)
install_dir: [String] Where filebeat should be installed (windows only)
tmp_dir: [String] Where filebeat should be temporarily downloaded to so it can be installed (windows only)
shutdown_timeout: [String] How long filebeat waits on shutdown for the publisher to finish sending events
beat_name: [String] The name of the beat shipper (default: FQDN)
tags: [Array] A list of tags that will be included with each published transaction
max_procs: [Number] The maximum number of CPUs that can be simultaneously used
fields: [Hash] Optional fields that should be added to each event output
fields_under_root: [Boolean] If set to true, custom fields are stored in the top level instead of under fields
disable_config_test: [Boolean] If set to true, configuration tests won't be run on config files before writing them.
processors: [Array] Processors that should be configured.
monitoring: [Hash] The monitoring.* components of the filebeat configuration.
inputs: [Hash] or [Array] Inputs that will be created. Commonly used to create inputs using hiera
setup: [Hash] Setup that will be created. Commonly used to create setup using hiera
xpack: [Hash] XPack configuration to pass to filebeat
extra_validate_options: [String] Extra command line options to pass to the configuration validation command.
overwrite_pipelines: [Boolean] If set to true, filebeat will overwrite existing pipelines.

Private Classes

Class: `filebeat::config`

Creates the configuration files required for filebeat (but not the inputs)

Class: `filebeat::install`

Calls the correct installer class based on the kernel fact.

Class: `filebeat::params`

Sets default parameters for filebeat based on the OS and other facts.

Class: `filebeat::repo`

Installs the yum or apt repository for the system package manager to install filebeat.

Class: `filebeat::service`

Configures and manages the filebeat service.

Class: `filebeat::install::linux`

Install the filebeat package on Linux kernels.

Class: `filebeat::install::windows`

Downloads, extracts, and installs the filebeat zip file in Windows.

Public Defines

Define: `filebeat::input`

Installs a configuration file for a input.

Be sure to read the filebeat configuration details to fully understand what these parameters do.

Parameters for filebeat::input

ensure: The ensure parameter on the input configuration file. (default: present)
paths: [Array] The paths, or blobs that should be handled by the input. (required if input_type is log)
containers_ids: [Array] If input_type is docker, the list of Docker container ids to read the logs from. (default: '*')
containers_path: [String] If input_type is docker, the path from where the logs should be read from. (default: /var/log/docker/containers)
containers_stream: [String] If input_type is docker, read from the specified stream only. (default: all)
combine_partial: [Boolean] If input_type is docker, enable partial messages joining. (default: false)
cri_parse_flags: [Boolean] If input_type is docker, enable CRI flags parsing from the log file. (default: false)
syslog_protocol: [Enum tcp,udp] Syslog protocol (default: udp)
syslog_host: [String] Host to listen for syslog messages (default: localhost:5140)
exclude_files: [Array] Files that match any regex in the list are excluded from filebeat (default: [])
encoding: [String] The file encoding. (default: plain)
input_type: [String] where filebeat reads the log from (default:log)
fields: [Hash] Optional fields to add information to the output (default: {})
fields_under_root: [Boolean] Should the fields parameter fields be stored at the top level of indexed documents.
ignore_older: [String] Files older than this field will be ignored by filebeat (default: ignore nothing)
close_older: [String] Files that haven't been modified since close_older, they'll be closed. New modifications will be read when files are scanned again according to scan_frequency. (default: 1h)
log_type: [String] (Deprecated - use doc_type) The document_type setting (optional - default: log)
doc_type: [String] The event type to used for published lines, used as type field in logstash and elasticsearch (optional - default: log)
scan_frequency: [String] How often should the input check for new files (default: 10s)
harvester_buffer_size: [Integer] The buffer size the harvester uses when fetching the file (default: 16384)
tail_files: [Boolean] If true, filebeat starts reading new files at the end instead of the beginning (default: false)
backoff: [String] How long filebeat should wait between scanning a file after reaching EOF (default: 1s)
max_backoff: [String] The maximum wait time to scan a file for new lines to ship (default: 10s)
backoff_factor: [Integer] backoff is multiplied by this parameter until max_backoff is reached to determine the actual backoff (default: 2)
force_close_files: [Boolean] Should filebeat forcibly close a file when renamed (default: false)
pipeline: [String] Filebeat can be configured for a different ingest pipeline for each input (default: undef)
include_lines: [Array] A list of regular expressions to match the lines that you want to include. Ignored if empty (default: [])
exclude_lines: [Array] A list of regular expressions to match the files that you want to exclude. Ignored if empty (default: [])
max_bytes: [Integer] The maximum number of bytes that a single log message can have (default: 10485760)
tags: [Array] A list of tags to send along with the log data.
json: [Hash] Options that control how filebeat handles decoding of log messages in JSON format See above. (default: {})
multiline: [Hash] Options that control how Filebeat handles log messages that span multiple lines. See above. (default: {})
host: [String] Host and port used to read events for TCP or UDP plugin (default: localhost:9000)
max_message_size: [String] The maximum size of the message received over TCP or UDP (default: undef)
keep_null: [Boolean] If this option is set to true, fields with null values will be published in the output document (default: undef)
include_matches: [Array] Journald input only, A collection of filter expressions used to match fields. The format of the expression is field=value (default: [])
seek: [Enum] Journald input only, The position to start reading the journal from (default: undef)
index: [String] If present, this formatted string overrides the index for events from this input (for elasticsearch outputs), or sets the raw_index field of the event’s metadata (for other outputs) (default: undef)
publisher_pipeline_disable_host: [Boolean] This disables the "host.name" attribute being added to events. See filebeat input configuration reference (default: false)

Limitations

This module doesn't load the elasticsearch index template into elasticsearch (required when shipping directly to elasticsearch).

When installing on Windows, there's an expectation that C:\Temp already exists, or an alternative location specified in the tmp_dir parameter exists and is writable by puppet. The temp directory is used to store the downloaded installer only.

Generic template

By default, a generic, open ended template is used that simply converts your configuration into a hash that is produced as YAML on the system. To use a template that is more strict, but possibly incomplete, set conf_template to filebeat/filebeat.yml.erb.

Debian Systems

Filebeat 5.x and newer requires apt-transport-https, but this module won't install it for you.

Using config_file

There are a few very specific use cases where you don't want this module to directly manage the filebeat configuration file, but you still want the configuration file on the system at a different location. Setting config_file will write the filebeat configuration file to an alternate location, but it will not update the init script. If you don't also manage the correct file (/etc/filebeat/filebeat.yml on Linux, C:/Program Files/Filebeat/filebeat.yml on Windows) then filebeat won't be able to start.

If you're copying the alternate config file location into the real location you'll need to include some metaparameters like

file { '/etc/filebeat/filebeat.yml':
  ensure  => file,
  source  => 'file:///etc/filebeat/filebeat.special',
  require => File['filebeat.yml'],
  notify  => Service['filebeat'],
}

to ensure that services are managed like you might expect.

Logging on systems with Systemd and with version filebeat 7.0+ installed

With filebeat version 7+ running on systems with systemd, the filebeat systemd service file contains a default that will ignore the logging hash parameter

Environment="BEAT_LOG_OPTS=-e`

to overide this default, you will need to set the systemd_beat_log_opts_override parameter to empty string

example:

class {'filebeat':
  logging => {
    'level'     => 'debug',
    'to_syslog' => false,
    'to_files'  => true,
    'files'     => {
      'path'        => '/var/log/filebeat',
      'name'        => 'filebeat',
      'keepfiles'   => '7',
      'permissions' => '0644'
    },
  systemd_beat_log_opts_override => "",
}

this will only work on systems with puppet version 6.1+. On systems with puppet version < 6.1 you will need to systemctl daemon-reload. This can be achived by using the camptocamp-systemd

include systemd::systemctl::daemon_reload

class {'filebeat':
  logging => {
...
    },
  systemd_beat_log_opts_override => "",
  notify  => Class['systemd::systemctl::daemon_reload'],
}

Development

Pull requests and bug reports are welcome. If you're sending a pull request, please consider writing tests if applicable.

sandwitch/puppet-filebeat