When doing a UPI install, the cluster will report as having gone through the IPI installation flow if the OPENSHIFT_INSTALL_INVOKER environment variable is specified. This was the intended behavior from https://bugzilla.redhat.com/show_bug.cgi?id=1791437, but we've since decided that we would like to distinguish between UPI and IPI more accurately. We need to follow up and change the way that the metrics count an IPI vs a UPI installation.
Reproduce this bug with 4.4.0-0.nightly-2020-02-04-171905. Trigger UPI install with OPENSHIFT_INSTALL_INVOKER ENV var setting. After installation, checking: # oc get cm -n openshift-config openshift-install -o yaml apiVersion: v1 data: invoker: jialiu version: v4.4.0 kind: ConfigMap metadata: creationTimestamp: "2020-02-06T08:57:36Z" name: openshift-install namespace: openshift-config resourceVersion: "963" selfLink: /api/v1/namespaces/openshift-config/configmaps/openshift-install uid: 8d32bd2f-ba75-45fd-bb79-642aa011228b # token=`oc sa get-token prometheus-k8s -n openshift-monitoring` # oc -n openshift-monitoring exec -c prometheus prometheus-k8s-0 -- curl -k -H "Authorization: Bearer $token" 'https://prometheus-k8s.openshift-monitoring.svc:9091/api/v1/query?query=cluster_installer' | jq % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 410 100 410 0 0 3487 0 --:--:-- --:--:-- --:--:-- 3504 { "status": "success", "data": { "resultType": "vector", "result": [ { "metric": { "__name__": "cluster_installer", "endpoint": "metrics", "instance": "10.0.65.75:9099", "invoker": "jialiu", "job": "cluster-version-operator", "namespace": "openshift-cluster-version", "pod": "cluster-version-operator-674c54969c-hkwz2", "service": "cluster-version-operator", "type": "openshift-install", -------> this line indicate it is a IPI install (if I am wrong, correct me) "version": "v4.4.0" }, "value": [ 1580998128.629, "1" ] } ] } }
Verified this bug with 4.4.0-0.nightly-2020-02-09-220310, and PASS. Whatever OPENSHIFT_INSTALL_INVOKER set or not, now openshift-install-manifests.yaml file always created when running `openshift-install create manifests` command # openshift-install create manifests --dir demo1 # cat demo1/openshift/openshift-install-manifests.yaml apiVersion: v1 data: invoker: user version: v4.4.0 kind: ConfigMap metadata: creationTimestamp: null name: openshift-install-manifests namespace: openshift-config OPENSHIFT_INSTALL_INVOKER value setting will also affect the `invoker` value in this yaml file. The difference between UPI and IPI is this configmap's name, in UPI it is "openshift-install-manifests", in IPI, this configmap's name is overwritten as "openshift-install". The configmap name will be used to determinate it is a upi or ipi in telemetry. Trigger a common IPI install without OPENSHIFT_INSTALL_INVOKER ENV. # oc get cm -n openshift-config openshift-install -o yaml apiVersion: v1 data: invoker: user version: v4.4.0 kind: ConfigMap metadata: creationTimestamp: "2020-02-10T02:20:27Z" name: openshift-install namespace: openshift-config resourceVersion: "1488" selfLink: /api/v1/namespaces/openshift-config/configmaps/openshift-install uid: 23d61d25-e41e-4dc7-8c88-c3490f871f5b # token=`oc sa get-token prometheus-k8s -n openshift-monitoring` # oc -n openshift-monitoring exec -c prometheus prometheus-k8s-0 -- curl -k -H "Authorization: Bearer $token" 'https://prometheus-k8s.openshift-monitoring.svc:9091/api/v1/query?query=cluster_installer' | jq % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 409 100 409 0 0 2543 0 --:--:-- --:--:-- --:--:-- 2540 { "status": "success", "data": { "resultType": "vector", "result": [ { "metric": { "__name__": "cluster_installer", "endpoint": "metrics", "instance": "10.0.128.86:9099", "invoker": "user", "job": "cluster-version-operator", "namespace": "openshift-cluster-version", "pod": "cluster-version-operator-54fd5685f7-5hnpz", "service": "cluster-version-operator", "type": "openshift-install", ===========> indicate this is ipi install "version": "v4.4.0" }, "value": [ 1581305053.812, "1" ] } ] } } Trigger a common IPI install without OPENSHIFT_INSTALL_INVOKER ENV. # oc get cm openshift-install-manifests -n openshift-config -o yaml apiVersion: v1 data: invoker: user version: v4.4.0 kind: ConfigMap metadata: creationTimestamp: "2020-02-10T03:29:22Z" name: openshift-install-manifests namespace: openshift-config resourceVersion: "1469" selfLink: /api/v1/namespaces/openshift-config/configmaps/openshift-install-manifests uid: 488f42ae-a2b9-4dbe-9041-5f0956dcc357 # token=`oc sa get-token prometheus-k8s -n openshift-monitoring` # oc -n openshift-monitoring exec -c prometheus prometheus-k8s-0 -- curl -k -H "Authorization: Bearer $token" 'https://prometheus-k8s.openshift-monitoring.svc:9091/api/v1/query?query=cluster_installer' | jq % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 344 100 344 0 0 2717 0 --:--:-- --:--:-- --:--:-- 2730 { "status": "success", "data": { "resultType": "vector", "result": [ { "metric": { "__name__": "cluster_installer", "endpoint": "metrics", "instance": "10.0.50.57:9099", "job": "cluster-version-operator", "namespace": "openshift-cluster-version", "pod": "cluster-version-operator-54fd5685f7-kl5lh", "service": "cluster-version-operator" }, "value": [ 1581307523.39, "1" ] } ] } } Trigger a common IPI install with OPENSHIFT_INSTALL_INVOKER ENV. # oc get cm -n openshift-config NAME DATA AGE admin-kubeconfig-client-ca 1 178m etcd-ca-bundle 1 178m etcd-metric-serving-ca 1 178m etcd-serving-ca 1 178m initial-etcd-ca 1 178m initial-kube-apiserver-server-ca 1 178m openshift-install-manifests 2 178m # oc get cm openshift-install-manifests -n openshift-config -o yaml apiVersion: v1 data: invoker: jialiu version: v4.4.0 kind: ConfigMap metadata: creationTimestamp: "2020-02-10T04:16:31Z" name: openshift-install-manifests namespace: openshift-config resourceVersion: "1401" selfLink: /api/v1/namespaces/openshift-config/configmaps/openshift-install-manifests uid: 864d06f4-eb6e-47ef-9a02-babfedcf9f12 # token=`oc sa get-token prometheus-k8s -n openshift-monitoring` # oc -n openshift-monitoring exec -c prometheus prometheus-k8s-0 -- curl -k -H "Authorization: Bearer $token" 'https://prometheus-k8s.openshift-monitoring.svc:9091/api/v1/query?query=cluster_installer' | jq % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 345 100 345 0 0 2436 0 --:--:-- --:--:-- --:--:-- 2446 { "status": "success", "data": { "resultType": "vector", "result": [ { "metric": { "__name__": "cluster_installer", "endpoint": "metrics", "instance": "10.0.55.97:9099", "job": "cluster-version-operator", "namespace": "openshift-cluster-version", "pod": "cluster-version-operator-54fd5685f7-2pzbf", "service": "cluster-version-operator" }, "value": [ 1581318899.596, "1" ] } ] } }
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:0581