elkstack

Elasticsearch, Logstash, and Kibana stack. Due to the recommendations of the community, we are not using the embedded elasticsearch functionality of logstash at this point. This cookbook provides recipes for all three components, along with wrapper recipes such as single or cluster to facilitate different use cases.

This stack's design is intended for one or many standalone nodes, with a full stack of elasticsearch, logstash, and kibana. The only difference between one and many nodes is that elasticsearch is clustered together. Data dispatched to Logstash on a particular node will use the local elasticsearch transport interface to index those logs to the node (and thus, the cluster). HTTP traffic dispatched to Kibana on port 80 on any node will also use the local elasticsearch HTTP interface to fetch and manipulate data.

Please read the individual recipe summaries to understand what each recipe does, as well as what each wrapper recipe is actually wrapping. As much as possible, upstream attributes have been exposed/overriden for our needs.

Things you should know

• This cookbook requires java. Because not everyone has the same desires for java versions, concurrently installed versions, or particular vendor versions, this cookbook simply assumes you have already satisfied this requirement. If you want just 'some java', feel free to use the ::java recipe and it will include the community java cookbook with default values.

• You must update your Berksfile to use this cookbook. Due to the upstream changes constantly occuring, you should consult the Berksfile in this cookbook and use its sources for kibana, logstash, and elasticsearch cookbooks. Eventually, as PRs get merged, this may no longer be a hard requirement. But the hardest thing will be that kibana in supermarket is currently a different actual cookbook.

• You should probably disable the nginx virtualhost that comes with the kibana cookbook and create your own configuration, securing it as appropriate for your own requirements. See the kibana_web LWRP documentation for more on what attributes should be set to accomplish this.

• If you'd like to disable backups using cloud files, set node['elkstack']['config']['backups']['enabled'] = false (it defaults to true). If you'd like to override the backup schedule/behavior for ES, simply disable the backup crontab entry by setting node['elkstack']['config']['backups']['cron']=false. This cookbook will still configure everything except the cronjob, and then you may create another one with your own schedule using the cron_d LWRP.

• Please note that this cookbook does not restart elasticsearch automatically, in order to avoid causing an outage of the cluster. It does restart nginx and logstash, however. You will have to restart elasticsearch after the initial bootstrap. You may also need to bounce logstash if it seems confused about losing a connection to eleasticsearch (unusual, but happens).

• You may want to consider adjusting node['elasticsearch']['discovery']['search_query'] if you are sharing one cluster among multiple environments. Just put a chef search in that attribute and this will use that search instead of one scoped to chef environments.

• You may want to consider adjusting node['elasticsearch']['allocated_memory'] if you are seeing an initial convergence failure (see #50). The chef client has been known to take up to 500mb or more on initial convergence. Combined with an initial allocation of 40% memory for ES, and 20% for logstash, that only leaves about 40% for the OS and chef. On a 2gb server, that ends up being 800mb for ES, about 400mb for logstash, leaving 800mb for the OS and the initial chef client run. After the initial run, memory footprint for the chef-client tends to be much, much lower, and ES is able to start.

• The agent and logstash recipes requires a pre-generated SSL key and certificate due to the requirements of the lumberjack protocol. This cookbook will consult node['elkstack']['config']['lumberjack_data_bag'] in order to locate and load a database that stores this key. It will first try an encrypted data bag, and if that doesn't work, will try an unencrypted data bag of the same name. If no data bag is found, it will autogenerate one and save it as an encrypted data bag. This means you must already have a 'secret file' on the node for an encryption key, as this is a require to use any encrypted data bags. To generate a key of your own, use something like:

openssl req -x509 -newkey rsa:2048 -keyout lumberjack.key -out lumberjack.crt -nodes -days 1000

This key and certificate data should be placed in data bag with name node['elkstack']['config']['lumberjack_data_bag'] under key and certificate keys, and base64 encoded into a single line string. You may also supply these secrets with some other method and populate the appropriate node.run_state values (see _secrets.rb for more details). Note that this is not a PKI trust model, but an explicit trust model.

There exists a make-lumberjack-key.sh to help you make this.

Changelog

See CHANGELOG.md for additional information about changes to this stack over time.

Supported Platforms

Ubuntu 12.04

Ubuntu 14.04

CentOS 6.5

Attributes

Customizing the stack

To override local storage for elasticsearch nodes (the stack will format and mount, as well as configure elasticsearch), set ['elkstack']['config']['data_disk']['disk_config_type'] to custom and provide each storage device and mount point in the following way:

disk_config = {
  'file_system' => 'ext4',
  'mount_options' => 'rw,user',
  'mount_path' => '/usr/local/var/data/elasticsearch/disk1',
  'format_command' => 'mkfs -t ext4 ',
  'fs_check_command' => 'dumpe2fs'
}

node.override['elasticsearch']['data']['devices']['/dev/xvde1'] = disk_config
node.override['elasticsearch']['path']['data'] = disk_config['mount_path']

To add additional logstash configuration to this stack, simply add additional templates in your wrapper cookbook. They should be placed in "#{@basedir}/#{@instance}/etc/conf.d" (see the config provider in the logstash cookbook). If you choose to use logstash-forwarder instead of the regular agent, please see the hash structure in attributes/forwarder.rb for adding additional files for the forwarder to watch and forward, node['logstash_forwarder']['config']['files'].

To override the nginx configuration, simply supply a new template and specify your cookbook using ['kibana']['nginx']['template_cookbook'] and ['kibana']['nginx']['template']. You can also override just the password for the reverse proxy using node.run_state['elkstack_kibana_password'].

By default, platformstack will call the ``::agentrecipe here. If you have a need for the forwarder recipe instead, just please note that you should turn off the platformstack flag for logging, and include theelkstack::agent` recipe directly.

To override anything else, set the appropriate node hash (logstash, kibana, or elasticsearch).

Usage

elkstack::default

Default recipe, does not do anything.

elkstack::single

A simple wrapper recipe that sets up Elasticsearch, Logstash, and Kibana. Also configures an rsyslog sink into logstash on the local box. Everything except Logstash and Kibana is locked down to listen only on localhost.

elkstack::cluster

A simple wrapper recipe that sets up Elasticsearch, Logstash, and Kibana. Also configures an rsyslog sink into logstash on the local box. Sets the cluster flag so that the elasticsearch recipe builds it in a cluster-friendly way.

elkstack::agent

A simple wrapper recipe that sets up a logstash agent on the local box. Also configures an rsyslog sink into logstash on the local box.

elkstack::forwarder

A python-based alternative to the normal agent, configured simply to watch logs forward them directly on to the cluster. This project is in heavy development, and is not publishing releases very often, so the packaged versions may be quite old or buggy. As of the addition of the recipe, the package was almost a year behind current development, but only because there also had been no releases either.

elkstack::elasticsearch

Leans on the upstream elasticsearch/cookbook-elasticsearch cookbook for much of its work. We do override the default set of plugins to be installed, as well as the amount of JVM heap. See attributes/default.rb for those settings.

This recipe also tags the node so that other nodes that run this recipe can discover it, and configure Elasticsearch appropriately to join their cluster. It uses a tag, the current chef environment, and the cluster name as the default search criteria.

Most of this is configurable using the upstream Elasticsearch cookbook's attributes, including the chef search itself. There is not an easy toggle to turn off the search, however. Enables iptables rules if default['elkstack']['iptables']['enabled'] not nil

elkstack::logstash

Leans on the upstream lusis/chef-logstash cookbook for much of its work. We do override the default set of plugins to be installed, as well as the amount of JVM heap. See attributes/default.rb for those settings.

elkstack::kibana

Leans on the upstream lusis/chef-kibana cookbook for most of its work. Sets up an nginx site for kibana by default. By default, it also does not pass through most of the http paths directly to elasticsearch (whitelist).

elkstack::java

Wrapper for a java recipe. This is not included on the run list normally, so if you don't already, you must include this recipe or get another JVM installed before including anything else in this cookbook.

elkstack::newrelic

Validates if there is a newrelic license set and based on that, see if the node is tagged as 'elkstack' or 'elkstack_cluster' and creates a file with elasticsearch details. Installs python, pip and setuptools packages in order to support newrelic_meetme_plugin

elkstack::acl

Adds cluster node basic iptables rules and cluster iptables rules if appropriate attributes are set.

elkstack::agent_acl

Adds agent node basic iptables rules if appropriate attributes are set.

elkstack::disk_setup

Look for node['elkstack']['config']['data_disk']['disk_config_type'] to be truthy, and configure the upstream elasticsearch cookbook to format, mount, and use devices appropriately.

elkstack::*_monitoring

These correspond with the recipes above, and just provide a way to pull out the monitoring work to make the original recipes cleaner.

Miscellaneous

The wrapper recipes are single and cluster. These change attributes and then invoke elasticsearch, logstash, kibana, and rsyslog. Finally, there are utility recipes like java and newrelic (not invoked otherwise), as well as acl which is called by _base if node['elkstack']['iptables']['enabled'].

Contributing

See CONTRIBUTING.

Authors

Author:: Rackspace ([email protected])

License

# Copyright 2014, Rackspace Hosting
#
# 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.
#