Skip to content

monitoring: support multi-bound alerts, add provisioning warnings for containers

Administrator requested to merge distribution/provisioning-alerts into master

Created by: bobheadxi

closes #7529 , closes #11237

Looks like upper-bound warnings are already in place - this PR adds some lower-bound warnings to round it off

Also adds support for multi-bound warnings to define an upper and lower bound (corresponding generated docs)

  - record: alert_count
    labels:
      description: 'frontend: 90%+ or less than 5% container memory usage by instance
        (not available on server)'
      level: warning
      name: container_memory_usage_low
      service_name: frontend
    expr: |-
      clamp_max(clamp_min(floor(
      (((5 / clamp_min(cadvisor_container_memory_usage_percentage_total{name=~".*frontend.*"}, 0.0000001)) OR on() vector(0)) >= 0) OR on() vector(1)
      ), 0), 1) OR on() vector(1)
  - record: alert_count
    labels:
      description: 'frontend: 90%+ or less than 5% container cpu usage total (5m average)
        across all cores by instance (not available on server)'
      level: warning
      name: container_cpu_usage_high
      service_name: frontend
    expr: |-
      clamp_max(clamp_min(floor(
      ((((cadvisor_container_cpu_usage_percentage_total{name=~".*frontend.*"}) / 90) OR on() vector(0)) >= 0) OR on() vector(1)
      ), 0), 1) OR on() vector(1)

Also updates some container monitoring text I missed in #10993

Merge request reports