Metrics

List of signals that Voithos uses for autoscaling

Container metrics

The Voithos collector scrapes the following container metrics from cadvisor.

container:
- container_cpu_usage_seconds_total
- container_cpu_cfs_periods_total
- container_cpu_cfs_throttled_periods_total
- container_memory_usage_bytes
- container_memory_working_set_bytes
- container_memory_failures_total

Kube state metrics

The Voithos controller scrapes metrics from the following fields of the kubernetes resources it autoscales.

kubeState:
- field: spec.replicas
  kinds:
  - deployment
  - replicaSet
  - statefulset
- field: status.replicas
  kinds:
  - deployment
  - replicaSet
  - statefulset
- field: status.numberAvailable
  kinds:
  - daemonset
- field: status.active
  kinds:
  - job
- field: spec.template.spec.containers[*].resources
  kinds:
  - daemonset
  - deployment
  - replicaSet
  - statefulset
  - job
- field: spec.jobTemplate.spec.template.spec.containers[*].resources
  kinds:
  - cronJob
- field: spec.template.spec.initContainers[*].resources
  kinds:
  - daemonset
  - deployment
  - replicaSet
  - statefulset
  - job
- field: spec.jobTemplate.spec.template.spec.initContainers[*].resources
  kinds:
  - cronjob