The core platform type is a key metric for debugging and triaging issues. It is currently not available on cluster, nor reported via telemetry. After discussion with apiserver team we will add cluster_infrastructure_provider{type="AWS",region="us-east-1"} 1 cluster_infrastructure_provider{type="Azure",region=""} 1 cluster_infrastructure_provider{type="GCP",region="us-central"} 1 cluster_infrastructure_provider{type="OpenStack",region=""} 1 cluster_infrastructure_provider{type="BareMetal",region=""} 1 cluster_infrastructure_provider{type="",region=""} 1 reporting from the kube-apiserver and then add that telemetry, allowing cluster type to be unambiguously determined from metrics and alerting.
Intended to be back ported to 4.1.
Verified in 4.2.0-0.nightly-2019-07-24-000310 AWS env, query cluster_infrastructure_provider in Prometheus UI, got 1 record: cluster_infrastructure_provider{endpoint="https",instance="10.128.0.63:8443",job="metrics",namespace="openshift-kube-apiserver-operator",pod="kube-apiserver-operator-84989775cf-dbzpg",region="us-east-2",service="metrics",type="AWS"} 1 It shows type and region well from kas operator. I think can move bug to VERIFIED.
Today checked Azure env, the metrics work well too: cluster_infrastructure_provider{endpoint="https",instance="10.130.0.11:8443",job="metrics",namespace="openshift-kube-apiserver-operator",pod="kube-apiserver-operator-65ff4f566c-8lh2c",service="metrics",type="Azure"} 1