VictoriaMetrics, Thanos, Prometheus

• Source file config
• Pipeline file config

Source file config

Property	Type	Description
type	String	Specify source type: victoria, prometheus, thanos
name	String	Unique source name - also the config file name
config	Object	Source configuration

All properties are required

config object properties:

Property	Type	Required	Description
url	String	yes	URL to fetch data from, e.g. http://victoriametrics:8428
username	String	no	username
password	String	no	password
verify_ssl	Boolean	no	whether to verify SSL certificate when connecting via HTTPS, default: true
query_timeout	Integer	no	Query timeout in seconds, 15 by default

Example

[
  {
  "name": "test_victoria",
  "type": "victoria",
  "config": {
    "url": "http://victoriametrics:8428",
    "username": "",
    "password": ""
  }
}
]

Example

[
  {
  "name": "prometheus_server",
  "type": "prometheus",
  "config": {
    "url": "http://prometheus_server:8428",
    "username": "",
    "password": ""
  }
}
]

Pipeline File config

Properties list

Property name in config file	Required	Value type in config file	Description
source	yes	String	Source config name
pipeline_id	yes	String	Unique pipeline identifier (use human-readable name so you could easily use it further)
query	yes	String	Query to export data, query may contain any PromQL selector
aggregated_metric_name	no	String	The name that will be applied to query results that aggregate multiple metrics (for example increase() or rate() functions)
days_to_backfill	yes	Integer	Collect data starting N days ago
interval	yes	Integer	Query data every N seconds
delay	no	Integer	Collect data with a specified delay, e.g. the pipeline will retrieve data until now -(minus) delay, it will not fetch the latest data immediately, unit - minutes
properties	no	Object with key-value pairs	Dimensions with static values to pass to Anodot.
tags	no	Object with key-value pairs	Tags
dimensions	no	List of dimensions	List of dimensions to extract from query results, needed only when using metric 3.0 protocol
values	no	List of metric names	List of metric names to extract from query results, needed only when using metric 3.0 protocol. Note: VictoriaMetrics supports only one metric per query.
units	no	Object	Key-value pairs (value:unit). The value must be from the values column, units can be any.
uses_schema	no	Boolean	Whether the pipeline uses schema or not
transform	no	Object	Object containing either transformation file path or transformation configuration, see transformations page for more details
request_headers	no	Object	Key-value pairs of custom headers to be used in queries
tag_configurations	no	Object	Configure tags with dynamic values. Each value in these configurations is a field. To learn about fields visit the fields wiki page

Examples

Simple:

[{
  "query": "{__name__=~\"go_gc_.*\"}",
  "days_to_backfill": 3,
  "interval": 10,
  "pipeline_id": "victoria_test",
  "source": "victoria_test"
}]

Advanced:

[{
  "query": "{__name__=~\"go_gc_.*\"}",
  "days_to_backfill": 3,
  "interval": 10,
  "pipeline_id": "victoria_test",
  "source": "victoria_test",
  "aggregated_metric_name": "my_metric",
  "delay": 10,
  "properties": { "version": "1.10" },
  "tags": { "env": "prod" },
  "request_headers": { "Host": "victoriametrics-prod1.com" }
}]

Using Protocol 3.0 and Anodot schema

To enable schema for PromQL metric pipelines you need to specify measurements and dimensions in the pipeline config. Note that only specified dimensions will be sent to Anodot, if the metric has other labels they will be ignored

  [{
    "source": "test_victoria",
    "pipeline_id": "test_promql_schema",
    "query": "diff_dims[10000s]",
    "days_to_backfill": 10,
    "interval": 10000,
    "dimensions": [
      "instance",
      "location",
      "job"
    ],
    "values": {
      "diff_dims": "gauge"
    },
    "units": {
      "diff_dims": "unit"
    },
    "tag_configurations": {
      "Tag_name1": {
        "value_path": "property1"
      },
      "Tag_name2": {
        "value_path": "property2"
      }
    }
  }]

If your query has aggregation functions like rate, you need to specify aggregated_metric_name in addition to dimension and values. The new name then has to be specified in the values config

  [{
    "source": "test_victoria",
    "pipeline_id": "test_promql_schema_rate",
    "query": "rate(diff_dims[10000s])",
    "days_to_backfill": 10,
    "interval": 10000,
    "aggregated_metric_name": "counter_rate",
    "dimensions": [
      "instance",
      "location",
      "job"
    ],
    "values": {
      "counter_rate": "gauge"
    },
    "units": {
      "counter_rate": "unit"
    }
  }]