Configure selective Prometheus metric collection

Limit Prometheus metrics by category, collection, or metric name.

Configure selective Prometheus metric collection

Limit exporter metrics with the collect[] and exclude[] parameters. Use these parameters with Prometheus jobs for Observe or Local WEKA Home.

Metric filters

collect[]: An allowlist that selects categories, collections, or individual metrics. Omit it to return all endpoint metrics.

exclude[]: A denylist that removes metrics from the selected set. It applies after collect[].

Categories and collections

A category groups metrics sharing a name prefix. For example, s3 returns metrics beginning weka_s3_. The drive category returns metrics beginning weka_drive_.

Categories include newly added metrics with their matching prefix. No configuration changes are needed.

Collections split large categories into focused subsets. For example, cluster_health returns health and alerting metrics without performance metrics.

Download the YAML metric reference from the exporter panel. It lists metrics, categories, and collections for your cluster version.

Configure metric collection

Add filter parameters to a Prometheus job.

Before you begin

• Configure a Prometheus scrape job for the desired exporter endpoint.

• Download the metric reference from the exporter panel.

Procedure

1. Select the category, collection, or metric names to collect.

2. Add them under params.collect[] in the scrape job.

3. Add metric names under params.exclude[] to remove them.

4. Reload Prometheus after saving the configuration.

Examples

Alerting metrics only

params:
  collect[]: [cluster_health]

Dashboard metrics

params:
  collect[]: [cluster_health, cluster_performance, s3, drive]

Exclude a high-cardinality metric

params:
  collect[]: [cluster_health, cluster_performance]
  exclude[]: [weka_cpu_utilization_percent]

Collect one metric

params:
  collect[]: [weka_cluster_status]

Complete alerting job

- job_name: weka_alerting
  scheme: https
  authorization:
    credentials: <your-token>
  metrics_path: <base-url>
  params:
    collect[]: [cluster_health]
  static_configs:
    - targets: ['<exporter-host>']

You can define several jobs for one endpoint. Use separate collect[] values for alerting and dashboards.