Prometheus configuration¶

The prometheus section in paas-config.yaml allows you to define custom Prometheus scrape targets for metrics collection. This configuration is useful when the default framework metrics are insufficient for your use case or you want to expose additional metrics endpoints.

The custom scrape configurations are merged with the default framework metrics job (if defined), so both can be active when you integrate with Prometheus.

Configuration schema¶

prometheus¶

The optional top-level prometheus section contains scrape configuration.

Field	Type	Description
`scrape_configs`	List	List of scrape job configurations. Optional.

scrape_configs¶

Each item in scrape_configs defines a Prometheus scrape job.

Field	Type	Description
`job_name`	String	Unique name for this scrape job. Required. Must be unique across all jobs.
`metrics_path`	String	HTTP path to scrape for metrics. Optional. Default: `/metrics`
`static_configs`	List	List of static target configurations. Required.

static_configs¶

Each static configuration defines a set of targets and optional labels.

Field	Type	Description
`targets`	List of strings	List of target endpoints to scrape. Required. Supports wildcards and placeholders.
`labels`	Dictionary	Key-value pairs of labels to attach to all metrics from these targets. Optional.

Target formats¶

The targets field supports several formats:

Wildcard targets¶

Use *:PORT to target all units in the application on the specified port:

targets:
  - "*:8081"  # Scrapes all units on port 8081

This target expands to the pod IP addresses of all units.

Scheduler-only targets¶

Use @scheduler:PORT to target only the scheduler services:

targets:
  - "@scheduler:8082"  # Scrapes only scheduler service on port 8082

Scheduler services are guaranteed to run in only one unit. See Worker and Scheduler Services.

The @scheduler placeholder resolves to the fully qualified domain name (FQDN) of the scheduler unit.

Specific hosts¶

You can also specify exact hostnames or IP addresses in the targets section. For example:

prometheus:
  scrape_configs:
    # Application metrics from all units
    - job_name: "flask-app-custom"
      metrics_path: "/metrics"
      static_configs:
        - targets:
            - "*:8081"
          labels:
            app: "flask"
            env: "example"

    # Scheduler-specific metrics
    - job_name: "flask-scheduler-metrics"
      metrics_path: "/metrics"
      static_configs:
        - targets:
            - "@scheduler:8082"
          labels:
            role: "scheduler"

Validation rules¶

The Prometheus configuration validates the following rules:

No extra fields are allowed in the schema.
Each job_name must be unique across all scrape configs.
Targets using the @scheduler:PORT format will require a numeric port.