Prometheus Connector

Introduction

This page covers prometheus connector properties for dataSource configuration and the nuances associated with prometheus connector.

Prometheus Connector Properties in DataSource Configuration

Prometheus Connector Properties.

prometheus.uri [Required].
- This parameters provides the URI information to connect to a prometheus instance.
prometheus.auth.type [Optional]
- This parameters provides the authentication type information.
- Prometheus connector currently supports basicauth and awssigv4 authentication mechanisms.
- If prometheus.auth.type is basicauth, following are required parameters.
  - prometheus.auth.username and prometheus.auth.password.
- If prometheus.auth.type is awssigv4, following are required parameters.
  - prometheus.auth.region, prometheus.auth.access_key and prometheus.auth.secret_key

Example prometheus dataSource configuration with different authentications

No Auth

[{
    "name" : "my_prometheus",
    "connector": "prometheus",
    "properties" : {
        "prometheus.uri" : "http://localhost:9090"
    }
}]

Basic Auth

[{
    "name" : "my_prometheus",
    "connector": "prometheus",
    "properties" : {
        "prometheus.uri" : "http://localhost:9090",
        "prometheus.auth.type" : "basicauth",
        "prometheus.auth.username" : "admin",
        "prometheus.auth.password" : "admin"
    }
}]

AWSSigV4 Auth

[{
    "name" : "my_prometheus",
    "connector": "prometheus",
    "properties" : {
        "prometheus.uri" : "http://localhost:8080",
        "prometheus.auth.type" : "awssigv4",
        "prometheus.auth.region" : "us-east-1",
        "prometheus.auth.access_key" : "{{accessKey}}",
        "prometheus.auth.secret_key" : "{{secretKey}}"
    }
}]

PPL Query support for prometheus connector

Metric as a Table

Each connector has to abstract the underlying datasource constructs into a table as part of the interface contract with the PPL query engine. Prometheus connector abstracts each metric as a table and the columns of this table are @value, @timestamp, label1, label2---. @value represents metric measurement and @timestamp represents the timestamp at which the metric is collected. labels are tags associated with metric queried. For eg: handler, code, instance, code are the labels associated with prometheus_http_requests_total metric. With this abstraction, we can query prometheus data using PPL syntax similar to opensearch indices. Sample Example

source = my_prometheus.prometheus_http_requests_total

Expected output:

+--------+-----------------------+--------------+------+------------+------------+
| @value | @timestamp            | handler      | code | instance   | job        |
|--------+-----------------------+--------------+------+------------+------------|
| 5      | "2022-11-03 07:18:14" | "/-/ready"   | 200  | 192.15.1.1 | prometheus |
| 3      | "2022-11-03 07:18:24" | "/-/ready"   | 200  | 192.15.1.1 | prometheus |
| 7      | "2022-11-03 07:18:34" | "/-/ready"   | 200  | 192.15.1.1 | prometheus |
| 2      | "2022-11-03 07:18:44" | "/-/ready"   | 400  | 192.15.2.1 | prometheus |
| 9      | "2022-11-03 07:18:54" | "/-/promql"  | 400  | 192.15.2.1 | prometheus |
| 11     | "2022-11-03 07:18:64" | "/-/metrics" | 500  | 192.15.2.1 | prometheus |
+--------+-----------------------+--------------+------+------------+------------+

Default time range and resolution

Since time range and resolution are required parameters for query apis and these parameters are determined in the following manner from the PPL commands.

Time range is determined through filter clause on @timestamp. If there is no such filter clause, time range will be set to 1h with endtime set to now().
In case of stats, resolution is determined by span(@timestamp,15s) expression. For normal select queries, resolution is auto determined from the time range set.