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.

<source>
  @type tail
  # ...
  <parse>
    # ...
  </parse>
</source>

Parser Plugin Type

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

<parse>
  @type apache2
</parse>

Third-party plugins may also be installed and configured.

For more details, see plugins documentation.

Parameters

@type

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

<parse>
  @type regexp
  # ...
</parse>

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

• ​regexp​

• ​apache2​

• ​apache_error​

• ​nginx​

• ​syslog​

• ​csv​

• ​tsv​

• ​ltsv​

• ​json​

• ​multiline​

• ​none​

Parse Parameters

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 detail of 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 (string) (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

  Fluent::EventTime.now(current time) as a timestamp when time_key is

  specified.

  • Default: true

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

  record.

  • 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

  type

  • Default: float

  • Available values: float, unixtime, string, mixed

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

      1510544836.154709804)

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

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

      zone

    • 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

      ​Time#strftime.

    • For more details about parsing, see

      ​Time.strptime.

    • %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.