DogStatsD Mapper
Datadog の調査レポート: サーバーレスの状態 レポート: サーバーレスの状態

DogStatsD Mapper

このページは英語では対応しておりません。随時翻訳に取り組んでいます。翻訳に関してご質問やご意見ございましたら、お気軽にご連絡ください。

With Agent v7.17+, the DogStatsD Mapper feature allows you to convert parts of a metric name submitted to DogStatsD to tags using mapping rules with wildcard and regex patterns. For example it allows you to transform the metric:

  • airflow.job.duration.<JOB_TYPE>.<JOB_NAME>

into the metric airflow.job.duration with two associated tags:

  • job_type:<JOB_TYPE>
  • job_name:<JOB_NAME>.

To create a mapping rule:

  1. Open your datadog.yaml file.
  2. Add a mapping rule configuration block under the dogstatsd_mapper_profiles parameter.

Mapping rule configuration

A mapping rule block has the following layout:

dogstatsd_mapper_profiles:
    - name: '<PROFILE_NAME>'
      prefix: '<PROFILE_PREFIX>'
      mappings:
          - match: '<METRIC_TO_MATCH>'
            match_type: '<MATCH_TYPE>'
            name: '<MAPPED_METRIC_NAME>'
            tags:
                '<TAG_KEY>': '<TAG_VALUE_TO_EXPAND>'

With the following placeholders:

Placeholder DefinitionRequired
 <PROFILE_NAME>A name to give to your mapping rule profile.yes 
<PROFILE_PREFIX> The metric name prefix associated to this profile.yes 
<METRIC_TO_MATCH> The metric name to extract groups from with the Wildcard or Regex match logic. yes
<MATCH_TYPE>The type of match to apply to the <METRIC_TO_MATCH>. Either wildcard or regex   no, default: wildcard
<MAPPED_METRIC_NAME>The new metric name to send to Datadog with the tags defined in the same group.yes
<TAG_KEY>The Tag key to associate to the tags collected. no
<TAG_VALUE_TO_EXPAND>The tags collected from the <MATCH_TYPE> to inline.no

Wildcard Match Pattern

The wildcard match pattern matches dot-separated metric names using * as wildcard. The metric name must be only composed of alphanumeric, ., and _ characters for this pattern to work. Groups extracted can then be expanded with the $n format e.g. $1, $2, $3… or the ${n} format e.g. ${1}, ${2}, ${3}, …

For instance, if you have the metric custom_metric.process.value_1.value_2 with the following mapping group configuration:

dogstatsd_mapper_profiles:
    - name: my_custom_metric_profile
      prefix: custom_metric.
      mappings:
          - match: 'custom_metric.process.*.*'
            match_type: wildcard
            name: custom_metric.process
            tags:
                tag_key_1: '$1'
                tag_key_2: '$2'

It would send the metric custom_metric.process to Datadog with the tags tag_key_1:value_1 and tag_key_2:value_2.

Note: If the incoming metric have already at least one tag, no mapping is applied.

Regex Match Pattern

The regex match pattern matches metric names using regex patterns. Compared to the wildcard match pattern, it allows to define captured groups that contain .. Groups extracted can then be expanded with the $n format e.g. $1, $2, $3… or the ${n} format e.g. ${1}, ${2}, ${3}, …

For instance, if you have the metric custom_metric.process.value_1.value.with.dots._2 with the following mapping group configuration:

dogstatsd_mapper_profiles:
    - name: my_custom_metric_profile
      prefix: custom_metric.
      mappings:
          - match: 'custom_metric\.process\.([\w_]+)\.(.+)'
            match_type: regex
            name: custom_metric.process
            tags:
                tag_key_1: '$1'
                tag_key_2: '$2'

It would send the metric custom_metric.process to Datadog with the tags tag_key_1:value_1 and tag_key_2:value.with.dots._2.

Note: If the incoming metric have already at least one tag, no mapping is applied.

Expand group in metric name

For the regex and wildcard match type, group collected can be expanded as tags value with an associated tag key as see above, but can also be used in the metric name parameter. For instance, if you have the metric custom_metric.process.value_1.value_2 with the following mapping group configuration:

dogstatsd_mapper_profiles:
    - name: my_custom_metric_profile
      prefix: custom_metric.
      mappings:
          - match: 'custom_metric.process.*.*'
            match_type: wildcard
            name: 'custom_metric.process.prod.$1.live'
            tags:
                tag_key_2: '$2'

It would send the metric custom_metric.process.prod.value_1.live to Datadog with the tag tag_key_2:value_2.

Further Reading