Versions | v1.0 (td-agent3)

Elasticsearch Output Plugin

The out_elasticsearch Output plugin writes records into Elasticsearch. By default, it creates records by bulk write operation. This means that when you first import records using the plugin, no record is created immediately.

The record will be created when the chunk_keys condition has been met. To change the output frequency, please specify the time in chunk_keys and specify timekey value in conf.

This document doesn't describe all parameters. If you want to know full features, check the Further Reading section.

Table of Contents

Installation

Since out_elasticsearch has been included in the standard distribution of td-agent since v3.0.1, td-agent users do not need to install it manually.

If you have installed Fluentd without td-agent, please install this plugin using fluent-gem.

$ fluent-gem install fluent-plugin-elasticsearch

Example Configuration

The following is a simple working configuration. This should serve as a good starting point for most users.

<match my.logs>
  @type elasticsearch
  host localhost
  port 9200
  logstash_format true
</match>

For more details on each option, read the section on Parameters.

Plugin helpers

Parameters

@type (required)

This option must be always elasticsearch.

host (optional)

The hostname of your Elasticsearch node (default: localhost).

port (optional)

The port number of your Elasticsearch node (default: 9200).

hosts (optional)

If you want to connect to more than one Elasticsearch nodes, specify this option in the following format:

hosts host1:port1,host2:port2,host3:port3
# or
hosts https://customhost.com:443/path,https://username:password@host-failover.com:443

If you use this option, the host and port options are ignored.

user, password (optional)

The login credentials to connect to the Elasticsearch node (default: nil)

user fluent
password mysecret

scheme (optional)

Specify https if your Elasticsearch endpoint supports SSL (default: http)

path (optional)

The REST API endpoint of Elasticsearch to post write requests (default: nil)

index_name (optional)

The index name to write events to (default: fluentd).

This option supports the placeholder syntax of Fluentd plugin API. For example, if you want to partition the index by tags, you can specify as below:

index_name fluentd.${tag}

Here is a more practical example which partitions the Elasticsearch index by tags and timestamps:

index_name fluentd.${tag}.%Y%m%d

Time placeholder needs to set up tag and time in chunk_keys. Also it needs to specify timekey for time slice of chunk:

<buffer tag, time>
  timekey 1h # chunks per hours ("3600" also available)
</buffer>

logstash_format (optional)

With this option set true, Fluentd uses the conventional index name format logstash-%Y.%m.%d (default: false). This option supersedes the index_name option.

@log_level option

The @log_level option allows the user to set different levels of logging for each plugin. The supported log levels are: fatal, error, warn, info, debug, and trace.

Please see the logging article for further details.

Miscellaneous

You can use %{} style placeholders to escape for URL encoding needed characters.

user %{demo+}
password %{@secret}

are valid configuration.

hosts https://%{j+hn}:%{passw@rd}@host1:443/elastic/,http://host2

are also valid configuration.

But,

user demo+
password @secret

are invalid configuration.

Common Output / Buffer parameters

For common output / buffer parameters, please check the following articles.

Troubleshooting

Please refer to the elasticsearch README’s troubleshooting section.

Further Reading

Last updated: 2018-11-16 17:17:21 +0000

Versions | v1.0 (td-agent3)

If this article is incorrect or outdated, or omits critical information, please let us know. Fluentd is a open source project under Cloud Native Computing Foundation (CNCF). All components are available under the Apache 2 License.