Hello, we would like to scale Kubernetes deployments based on custom metrics from namespace-local Prometheus servers, in other words, multiple (tens) Prometheus servers per cluster. Has this been done?
As far as I can tell, the custom metrics API <https://github.com/kubernetes/community/blob/master/contributors/design-proposals/instrumentation/custom-metrics-api.md#api-paths> specifies a specific endpoint <https://github.com/kubernetes/metrics/blob/21468940b5195da44f899b8104af96e45d383f3f/pkg/apis/custom_metrics/register.go#L25>, so that I can only specify one ApiService to one Prometheus adapter <https://github.com/openshift/k8s-prometheus-adapter>, which in turn only accepts one Prometheus URL <https://github.com/openshift/k8s-prometheus-adapter/blob/de2e79e2b5243b1d4eb77969a8f30fbe450fffc1/cmd/adapter/adapter.go#L111-L112>. Currently, we work around this by federating selected metrics to one Prometheus server per cluster; however this introduces additional delay in the autoscaling control loop. Is there another way? I noticed that KEDA <https://keda.sh> allows Prometheus triggers that each specify a Prometheus URL <https://keda.sh/docs/1.5/scalers/prometheus/>. Is this useful for continuous scaling based on metrics such as the (time-averaged) number of concurrent requests in flight? Any experiences appreciated! Best, MR -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/CAFU3N5WQb6rWyvutiGw9svAurVpv4wGyJ_dSQgM66a%3DS_njtfQ%40mail.gmail.com.
