Skip to main content

OTEL-054: Prometheus scrape_interval below 10s

Severity: warn (advisory)

Rule Details​

Most Prometheus metrics have sample jitter that makes sub-10s scraping statistically noisy. What you do get is a lot more load on the target (Go runtime metrics in particular are expensive to serialize) and a lot more storage in the backend. 10s is a safe baseline; go below it only with a clear need and data to back it up.

This rule fires when any prometheus scrape config has scrape_interval < 10s.

Options​

FieldConstraint
scrape_intervalShould be ≥ 10s

Examples​

Avoid
receivers:
  prometheus:
    config:
      scrape_configs:
        - job_name: app
          scrape_interval: 1s       # too aggressive
          static_configs:
            - targets: [app:9090]
Prefer
receivers:
  prometheus:
    config:
      scrape_configs:
        - job_name: app
          scrape_interval: 15s
          static_configs:
            - targets: [app:9090]

When Not To Use It​

Short-lived diagnostic runs where you want high-frequency samples for a few minutes. Use a separate temporary job in that case and leave the rule enabled globally.

  • OTEL-055 — hostmetrics collection_interval below 10s

Version​

Available since augur v0.1.0.

Further Reading​

Resources​