Prometheus Adapter
Wrapper around the upstream prometheus-adapter Helm
chart (v5.3.0). Implements the Kubernetes custom.metrics.k8s.io and external.metrics.k8s.io APIs by
translating PromQL queries against Prometheus into metrics the API server (and HPA) can consume.
Why it's in KubeAid
Enables Horizontal Pod Autoscaling on metrics other than CPU/memory - request rate, queue depth, or any
other Prometheus series - by exposing them through the custom metrics API. See the autoscaling section of
Prometheus Configuration, which documents adding rules via
prometheus_adapter_additional_rules in the cluster's Jsonnet config and enabling
enable_custom_metrics_apiservice.
Prerequisites
kube-prometheus(Prometheus) running in-cluster - this chart points at it directly.
Key values / KubeAid-specific configuration
| Value | Description | Default |
|---|---|---|
prometheus-adapter.prometheus.url | Prometheus service the adapter queries | kube-prometheus-stack-prometheus.monitoring.svc |
Custom metric rules themselves are not set in this chart's values.yaml - they are generated from the
cluster's kube-prometheus Jsonnet config (prometheus_adapter_additional_rules) and applied alongside the
rest of the kube-prometheus build.