<a name="title"></a> chef-logstash

Description

This is the semi-official 'all-in-one' Logstash cookbook.

This cookbook is in transition from being a regular cookbook to following the Library Cookbook pattern.
While you can still use the agent and server recipes, the power of this cookbook now comes from the
LWRPs.

If you are using logstash < 1.2 you might want to use the 0.6.x branch.
If you are using logstash < 1.4 you might want to use the 0.7.x branch.

Requirements

All of the requirements are explicitly defined in the recipes. Every
effort has been made to utilize Opscode's cookbooks.

However if you wish to use an external ElasticSearch cluster, you will
need to install that yourself and change the relevant attributes for
discovery. The same applies to integration with Graphite.

This cookbook has been tested together with the following cookbooks,
see the Berksfile for more details

• Heavywater Graphite Cookbook - This is the one I use
• Karmi's ElasticSearch Cookbook
• RiotGames RBENV cookbook
• @lusis Kibana cookbook

Attributes

Default

see [attributes/default.rb](attributes/default.rb)

Beaver (alternative to Logstash Agent)

This will be depreciated soon in favor of an external library cookbook.

• node['logstash']['beaver']['repo'] - URL or repository to install beaver from (using pip).
• node['logstash']['beaver']['server_role'] - The role of the node behaving as a Logstash server/indexer.
• node['logstash']['beaver']['server_ipaddress'] - Server IP address to use (needed when not using server_role).

• node['logstash']['beaver']['inputs'] - Array of input plugins
  configuration (Supported: file).
  For example:

  override['logstash']['beaver']['inputs'] =  [
    { :file =>  
      {
        :path => ["/var/log/nginx/*log"], 
        :type => "nginx", 
        :tags => ["logstash","nginx"]
      }
    },
    { :file =>  
      {
        :path => ["/var/log/syslog"], 
        :type => "syslog", 
        :tags => ["logstash","syslog"] 
      }
    }
  ]    
  

• node['logstash']['beaver']['outputs'] - Array of output plugins
  configuration (Supported: amq, redis, stdout, zeromq).
  For example:

  override['logstash']['beaver']['outputs'] = [ 
    { 
      :amqp => { 
        :port => "5672",
        :exchange => "rawlogs",
        :name => "rawlogs_consumer"
      } 
    } 
  ]
  

  This example sets up the amqp output and uses the recipe defaults for the host value

Source

• node['logstash']['source']['repo'] - The git repo to use for the source code of Logstash
• node['logstash']['source']['sha'] - The sha/branch/tag of the repo you wish to clone. Uses node['logstash']['server']['version'] by default.
• node['logstash']['source']['java_home'] - your JAVA_HOME location. Needed explicity for ant when building JRuby

Index Cleaner

• node['logstash']['index_cleaner']['days_to_keep'] - Integer number of days from today of Logstash index to keep.
• node['logstash']['index_cleaner']['cron']['minute'] - Minute to run the index_cleaner cron job
• node['logstash']['index_cleaner']['cron']['hour'] - Hour to run the index_cleaner cron job
• node['logstash']['index_cleaner']['cron']['log_file'] - Path to direct the index_cleaner cron job's stdout and stderr

Lightweight Resource Providers

These now do all the heavy lifting.

logstash_instance

This will install a logstash instance. It will take defaults from attributes for most attributes.

see [resources/instance.rb](resources/instance.rb)

logstash_service

This will create system init scripts for managing logstash instance. It will take defaults from attributes for most attributes.

see [resources/service.rb](resources/service.rb)

experimental support for pleaserun has been added. Only native for Ubuntu 12.04 has been thoroughly tested.

logstash_config

This will create logstash config files. It will take defaults from attributes for most attributes.

see [resources/config.rb](resources/config.rb)

logstash_pattern

This will install custom grok patterns for logstash. It will take defaults from attributes for most attributes:

see [resources/pattern.rb](resources/pattern.rb)

logstash_plugns

This will install the logstash community plugins:

see [resources/plugins.rb](resources/plugins.rb)

attribute precidence in logstash LWRPs

We've done our best to make this intuitive and easy to use.

1. the value directly in the resource call.
2. the value from the hash node['logstash']['instance'][name]
3. the value from the hash node['logstash']['instance']['default']

Searching

There is a search helper library libraries/search.rb which will help you search for values such as elasticsearch_ip. see the server recipe for an example of its usage.

Testing

Vagrant

vagrant up precise64

Rubocop, FoodCritic, Rspec, Test-Kitchen

bundle exec rake

Contributing

Any and all contributions are welcome. We do ask that you test your contributions with the testing framework before you send a PR.

Documentation contributions will earn you lots of hugs and kisses.

Usage

A proper readme is forthcoming but in the interim....

These two recipes show how to install and configure logstash instances via the provided LWRPs

• [recipe/server.rb](recipe/server.rb) - This would be your indexer node
• [recipe/agent.rb](recipe/agent.rb) - This would be a local host's agent for collection

Every attempt (and I mean this) was made to ensure that the following
objectives were met:

• Any agent install can talk to a server install
• Kibana web interface can talk to the server install
• Each component works OOB and with each other
• Utilize official opscode cookbooks where possible

This setup makes HEAVY use of roles. Additionally, ALL paths have been
made into attributes. Everything I could think of that would need to
be customized has been made an attribute.

Defaults

By default, the recipes look for the following roles (defined as
attributes so they can be overridden):

• graphite_server - node['logstash']['graphite_role']
• elasticsearch_server - node['logstash']['elasticsearch_role']
• logstash_server - node['logstash']['kibana']['elasticsearch_role'] and node['logstash']['agent']['server_role']

The reason for giving kibana its own role assignment is to allow you
to point to existing ES clusters/logstash installs.

The reason for giving agent its own role assignment is to allow the
server and agent recipes to work together.

Yes, if you have a graphite installation with a role of
graphite_server, logstash will send stats of events received to
logstash.events.

Agent and Server configuration

The template to use for configuration is made an attribute as well.
This allows you to define your OWN logstash configuration file without
mucking with the default templates.

The server will, by default, enable the embedded ES server. This can
be overriden as well.

See the server and agent attributes for more details.

Source vs. Jar install methods

Both agent and server support an attribute for how to install. By
default this is set to jar to use the 1.1.1preview as it is required
to use elasticsearch 0.19.4. The current release is defined in
attributes if you choose to go the source route.

Out of the box behaviour

Here are some basic steps

• Create a role called logstash_server and assign it the following recipes: logstash::server
• Assign the role to a new server
• Assign the logstash::agent recipe to another server

If there is a system found with the logstash_server role, the agent
will automatically configure itself to send logs to it over tcp port
5959. This is, not coincidently, the port used by the chef logstash
handler.

If there is NOT a system with the logstash_server role, the agent
will use a null output. The default input is to read files from
/var/log/*.log excluding and gzipped files.

If you point your browser to the logstash_server system's ip
address, you should get the kibana web interface.

Do something to generate a new line in any of the files in the agent's
watch path (I like to SSH to the host), and the events will start
showing up in kibana. You might have to issue a fresh empty search.

The pyshipper recipe will work as well but it is NOT wired up to
anything yet.

config templates

If you want to use chef templates to drive your configs you'll want to set the following:

• example using agent, server works the same way.
• The actual template file for the following would resolve to templates/default/apache.conf.erb and be installed to /opt/logstash/agent/etc/conf.d/apache.conf
• Each template has a hash named for it to inject variables in node['logstash']['agent']['config_templates_variables']

node['logstash']['agent']['config_file'] = "" # disable data drive templates ( can be left enabled if want both )
node['logstash']['agent']['config_templates'] = ["apache"]
node['logstash']['agent']['config_templates_cookbook'] = 'logstash'
node['logstash']['agent']['config_templates_variables'] = { apache: { type: 'apache' } }

Letting data drive your templates

DEPRECIATED!

While this may work ... it is no longer being actively supported by the maintainers of this cookbook.
We will accept PRs.

The current templates for the agent and server are written so that you
can provide ruby hashes in your roles that map to inputs, filters, and
outputs. Here is a role for logstash_server.

There are two formats for the hashes for filters and outputs that you should be aware of ...

Legacy

This is for logstash < 1.2.0 and uses the old pattern of setting 'type' and 'tags' in the plugin to determine if it should be run.

filters: [
  grok: {
  type: "syslog"
    match: [
      "message",
      "%{SYSLOGTIMESTAMP:timestamp} %{IPORHOST:host} (?:%{PROG:program}(?:\[%{POSINT:pid}\])?: )?%{GREEDYDATA:message}"
    ]
  },
  date: {
  type: "syslog"
    match: [ 
      "timestamp",
      "MMM  d HH:mm:ss",
      "MMM dd HH:mm:ss",
      "ISO8601"
    ]
  }
]

Conditional

This is for logstash >= 1.2.0 and uses the new pattern of conditioansl if 'type' == "foo" {}

Note: the condition applies to all plugins in the block hash in the same object.

filters: [
  { 
    condition: 'if [type] == "syslog"',
    block: {    
      grok: {
        match: [
          "message",
          "%{SYSLOGTIMESTAMP:timestamp} %{IPORHOST:host} (?:%{PROG:program}(?:\[%{POSINT:pid}\])?: )?%{GREEDYDATA:message}"
        ]
      },
      date: {
        match: [ 
          "timestamp",
          "MMM  d HH:mm:ss",
          "MMM dd HH:mm:ss",
          "ISO8601"
        ]
      }
    }
  }
]

Examples

These examples show the legacy format and need to be updated for logstash >= 1.2.0

name "logstash_server"
description "Attributes and run_lists specific to FAO's logstash instance"
default_attributes(
  :logstash => {
    :server => {
      :enable_embedded_es => false,
      :inputs => [
        :rabbitmq => {
          :type => "all",
          :host => "",
          :exchange => "rawlogs"
        }
      ],
      :filters => [
        :grok => {
          :type => "haproxy",
          :pattern => "%{HAPROXYHTTP}",
          :patterns_dir => '/opt/logstash/server/etc/patterns/'
        }
      ],
      :outputs => [
        :file => {
          :type => 'haproxy',
          :path => '/opt/logstash/server/haproxy_logs/%{request_header_host}.log',
          :message_format => '%{client_ip} - - [%{accept_date}] "%{http_request}" %{http_status_code} ....'
        }
      ]
    }
  }
)
run_list(
  "role[elasticsearch_server]",
  "recipe[logstash::server]"
)

It will produce the following logstash.conf file

input {

  amqp {
    exchange => 'rawlogs'
    host => ''
    name => 'rawlogs_consumer'
    type => 'all'
  }
}

filter {

  grok {
    pattern => '%{HAPROXYHTTP}'
    patterns_dir => '/opt/logstash/server/etc/patterns/'
    type => 'haproxy'
  }
}

output {
  stdout { debug => true debug_format => "json" }
  elasticsearch { host => "127.0.0.1" cluster => "logstash" }

  file {
    message_format => '%{client_ip} - - [%{accept_date}] "%{http_request}" %{http_status_code} ....'
    path => '/opt/logstash/server/haproxy_logs/%{request_header_host}.log'
    type => 'haproxy'
  }
}

Here is an example using multiple filters

default_attributes(
  :logstash => {
    :server => {
      :filters => [
        { :grep => {
            :type => 'tomcat',
            :match => { '@message' => '([Ee]xception|Failure:|Error:)' },
            :add_tag => 'exception',
            :drop => false
        } },
        { :grep => {
            :type => 'tomcat',
            :match => { '@message' => 'Unloading class ' },
            :add_tag => 'unloading-class',
            :drop => false
        } },
        { :multiline => {
            :type => 'tomcat',
            :pattern => '^\s',
            :what => 'previous'
        } }
      ]
    }
  }
)

It will produce the following logstash.conf file

filter {

  grep {
    add_tag => 'exception'
    drop => false
    match => ['@message', '([Ee]xception|Failure:|Error:)']
    type => 'tomcat'
  }

  grep {
    add_tag => 'unloading-class'
    drop => false
    match => ["@message", "Unloading class "]
    type => 'tomcat'
  }

  multiline {
    patterns_dir => '/opt/logstash/patterns'
    pattern => '^\s'
    type => 'tomcat'
    what => 'previous'
  }

}

Adding grok patterns

Grok pattern files can be generated using attributes as follows

default_attributes(
  :logstash => {
    :patterns => {
      :apache => {
        :HTTP_ERROR_DATE => '%{DAY} %{MONTH} %{MONTHDAY} %{TIME} %{YEAR}',
        :APACHE_LOG_LEVEL => '[A-Za-z][A-Za-z]+',
        :ERRORAPACHELOG => '^\[%{HTTP_ERROR_DATE:timestamp}\] \[%{APACHE_LOG_LEVEL:level}\](?: \[client %{IPORHOST:clientip}\])?',
      },
      :mywebapp => {
        :MYWEBAPP_LOG => '\[mywebapp\]',
      },
    },
    [...]
  }
)

This will generate the following files:

/opt/logstash/server/etc/patterns/apache

APACHE_LOG_LEVEL [A-Za-z][A-Za-z]+
ERRORAPACHELOG ^\[%{HTTP_ERROR_DATE:timestamp}\] \[%{APACHE_LOG_LEVEL:level}\](?: \[client %{IPORHOST:clientip}\])?
HTTP_ERROR_DATE %{DAY} %{MONTH} %{MONTHDAY} %{TIME} %{YEAR}

/opt/logstash/server/etc/patterns/mywebapp

MYWEBAPP_LOG \[mywebapp\]

This patterns will be included by default in the grok and multiline
filters.

Vagrant

Requirements

• Vagrant 1.2.1+
• Vagrant Berkshelf Plugin vagrant plugin install vagrant-berkshelf
• Vagrant Omnibus Plugin vagrant plugin install vagrant-omnibus

Uses the Box Name to determine the run list ( based on whether its Debian or RHEL based ).

See chef_json and chef_run_list variables to change recipe behavior.

Usage:

Run Logstash on Ubuntu Lucid : vagrant up lucid32 or vagrant up lucid64

Run Logstash on Centos 6 32bit : vagrant up centos6_32

Logstash will listen for syslog messages on tcp/5140

BIG WARNING

• Currently only tested on Ubuntu Natty, Precise, and RHEL 6.2.

License and Author

• Author: John E. Vincent
• Author: Bryan W. Berry (bryan.berry@gmail.com)
• Author: Richard Clamp (@richardc)
• Author: Juanje Ojeda (@juanje)
• Author: @benattar
• Author: Paul Czarkowski (@pczarkowski)
• Copyright: 2012, John E. Vincent
• Copyright: 2012, Bryan W. Berry
• Copyright: 2012, Richard Clamp
• Copyright: 2012, Juanje Ojeda
• Copyright: 2012, @benattar
• Copyright: 2014, Paul Czarkowski

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.