Config: Parse Section

Some of the Fluentd plugins support the <parse> section to specify how to parse the raw data.

Parse Section Overview

The parse section can be under <source>, <match> or <filter> section. It is enabled for the plugins that support parser plugin features.

  @type tail
  # ...
    # ...

Parser Plugin Type

The @type parameter of <parse> section specifies the type of the parser plugin. Fluentd core bundles some useful parser plugins.

  @type apache2

Third-party plugins may also be installed and configured.

For more details, see plugins documentation.



The @type parameter specifies the type of the parser plugin.

  @type regexp
  # ...

Here's the list of built-in parser plugins:

Parse Parameters

NOTE: It depends on an input plugin's capabilities whether can handle a <parse> section or not. See each plugin's documentation in detail.

The default value of the following parameters will be overridden by the individual parser plugins:

  • types (hash) (optional): Specify types for converting field into another

    type. See below the details of the "types Parameter" section.

    • Default: nil

    • string-based hash: field1:type, field2:type, field3:type:option, field4:type:option

    • JSON format: {"field1":"type", "field2":"type", "field3":"type:option", "field4":"type:option"}

    • example: types user_id:integer,paid:bool,paid_usd_amount:float

  • time_key (string) (optional): Specify time field for event time. If

    the event doesn't have this field, current time is used.

    • Default: nil

    • Note that json,

      ltsv and

      regexp override the default value of this

      parameter and set it to time by default.

  • null_value_pattern (regexp) (optional): Specify null value pattern.

    • Default: nil

  • null_empty_string (bool) (optional): If true, empty string field is

    replaced with nil.

    • Default: false

  • estimate_current_event (bool) (optional): If true, use time) as a timestamp when time_key is


    • Default: true

  • keep_time_key (bool) (optional): If true, keep time field in the


    • Default: false

  • timeout (time) (optional): Specify timeout for parse processing. This

    is mainly for detecting wrong regexp pattern.

    • Default: nil

types Parameter

For the types parameter, the following types are supported:

  • string: Converts the field into String type. This uses to_s method for conversion.

  • bool: Converts the string "true", "yes" or "1" into true. Otherwise, false.

  • integer (not int): Converts the field into the Integer type. This uses to_i method for conversion. For example, the string "1000" converts into 1000.

  • float: Converts the field into Float type. This uses to_f method for conversion. For example, the string "7.45" converts into 7.45.

  • time: Converts the field into Fluent::EventTime type. This uses Fluentd time parser for conversion. For the time type, the third field specifies the time format similar to time_format.

      date:time:%d/%b/%Y:%H:%M:%S %z # for string with time format
      date:time:unixtime             # for integer time
      date:time:float                # for float time

    See time_type and time_format parameters in Time parameters section.

  • array: Converts the string field into Array type. For the array type, the third field specifies the delimiter (the default is comma ","). For example, if a field item_ids contains the value "3,4,5", types item_ids:array parses it as ["3", "4", "5"]. Alternatively, if the value is "Adam|Alice|Bob", types item_ids:array:| parses it as ["Adam", "Alice", "Bob"].

Time Parameters

  • time_type (enum) (optional): parses/formats value according to this


    • Default: string

    • Available values: float, unixtime, string, mixed

      • float: seconds from Epoch + nano seconds (e.g.


      • unixtime: seconds from Epoch (e.g. 1510544815)

      • string: use format specified by time_format, local time or time


      • mixed: enable time_format_fallbacks option. (Since Fluentd v1.12.2)

  • time_format (string) (optional): processes value according to the

    specified format. This is available only when time_type is string.

    • Default: nil

    • Available time format:

      • For more details about formatting, see


      • For more details about parsing, see


      • %iso8601 (only for parsing)

      • Use %N to parse/format with sub-second precision, because

        strptime does not support

        %3N, %6N, %9N, and %L.

  • localtime (bool) (optional): if true, uses local time. Otherwise,

    UTC is used. This is exclusive with utc.

    • Default: true

  • utc (bool) (optional): if true, uses UTC. Otherwise, local time is

    used. This is exclusive with localtime.

    • Default: false

  • timezone (string) (optional): uses the specified timezone. One can

    parse/format the time value in the specified timezone format.

    • Default: nil

    • Available time zone format:

      1. [+-]HH:MM (e.g. "+09:00") (recommended)

      2. [+-]HHMM (e.g. "+0900")

      3. [+-]HH (e.g. "+09")

      4. Region/Zone (e.g. Asia/Tokyo)

      5. Region/Zone/Zone (e.g. America/Argentina/Buenos_Aires)

  • time_format_fallbacks () (optional): uses the specified time format as a fallback in the specified order.

    You can parse undetermined time format by using time_format_fallbacks. This options is enabled when time_type is mixed.

    • Default: nil

      time_type mixed
      time_format unixtime
      time_format_fallbacks %iso8601

    In the above use case, the timestamp is parsed as unixtime at first, if it fails, then it is parsed as %iso8601 secondary. Note that time_format_fallbacks is the last resort to parse mixed timestamp format. There is a performance penalty (Typically, N fallbacks are specified in time_format_fallbacks and if the last specified format is used as a fallback, N times slower in the worst case).

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

Last updated