https://bugzilla.redhat.com/show_bug.cgi?id=1952049#c9 identifies kcm as offender triggering the APIRemovedInNextReleaseInUse alert. This is wrong. kcm with its dynamic informers for quota, gc and namespace deletion has to be excluded from the alert. +++ This bug was initially created as a clone of Bug #1947719 +++ Created attachment 1770482 [details] alert screen shot Created attachment 1770482 [details] alert screen shot Description of problem: 8 DeprecatedAPIInUse info alerts display Version-Release number of selected component (if applicable): 4.8.0-0.nightly-2021-04-08-200632 How reproducible: always Steps to Reproduce: 1. open console-monitoring-alerts 2. 3. Actual results: 8 DeprecatedAPIInUse info alerts display Expected results: No other alerts display except watchdog Additional info: alert rule metrics: group by(group, version, resource) (apiserver_requested_deprecated_apis{removed_release="1.22"}) and (sum by(group, version, resource) (rate(apiserver_request_total[10m]))) > 0 Element Value: {group="rbac.authorization.k8s.io",resource="roles",version="v1beta1"} 1 {group="admissionregistration.k8s.io",resource="mutatingwebhookconfigurations",version="v1beta1"} 1 {group="admissionregistration.k8s.io",resource="validatingwebhookconfigurations",version="v1beta1"} 1 {group="apiextensions.k8s.io",resource="customresourcedefinitions",version="v1beta1"} 1 {group="certificates.k8s.io",resource="certificatesigningrequests",version="v1beta1"} 1 {group="extensions",resource="ingresses",version="v1beta1"} 1 {group="rbac.authorization.k8s.io",resource="clusterrolebindings",version="v1beta1"} 1 {group="rbac.authorization.k8s.io",resource="rolebindings",version="v1beta1"} 1 ---------------- # for i in roles mutatingwebhookconfigurations validatingwebhookconfigurations customresourcedefinitions certificatesigningrequests ingresses clusterrolebindings rolebindings; do oc api-resources | grep $i; echo -e "\n"; done clusterroles authorization.openshift.io/v1 false ClusterRole roles authorization.openshift.io/v1 true Role clusterroles rbac.authorization.k8s.io/v1 false ClusterRole roles rbac.authorization.k8s.io/v1 true Role mutatingwebhookconfigurations admissionregistration.k8s.io/v1 false MutatingWebhookConfiguration validatingwebhookconfigurations admissionregistration.k8s.io/v1 false ValidatingWebhookConfiguration customresourcedefinitions crd,crds apiextensions.k8s.io/v1 false CustomResourceDefinition certificatesigningrequests csr certificates.k8s.io/v1 false CertificateSigningRequest ingresses config.openshift.io/v1 false Ingress ingresses ing extensions/v1beta1 true Ingress ingresses ing networking.k8s.io/v1 true Ingress clusterrolebindings authorization.openshift.io/v1 false ClusterRoleBinding clusterrolebindings rbac.authorization.k8s.io/v1 false ClusterRoleBinding clusterrolebindings authorization.openshift.io/v1 false ClusterRoleBinding rolebindings authorization.openshift.io/v1 true RoleBinding clusterrolebindings rbac.authorization.k8s.io/v1 false ClusterRoleBinding rolebindings rbac.authorization.k8s.io/v1 true RoleBinding --- Additional comment from Junqi Zhao on 2021-04-09 05:28:56 CEST --- alert details alert:DeprecatedAPIInUse expr:group by(group, version, resource) (apiserver_requested_deprecated_apis{removed_release="1.22"}) and (sum by(group, version, resource) (rate(apiserver_request_total[10m]))) > 0 for: 1h labels: severity: info annotations: message: Deprecated API that will be removed in the next version is being used. Removing the workload that is using the {{"{{$labels.group}}"}}.{{"{{$labels.version}}"}}/{{"{{$labels.resource}}"}} API might be necessary for a successful upgrade to the next cluster version. Refer to the audit logs to identify the workload. --- Additional comment from hongyan li on 2021-04-09 05:37:17 CEST --- --- Additional comment from hongyan li on 2021-04-09 05:44:46 CEST --- Different issue from bug 1932165 which is about variable not translated to value --- Additional comment from Junqi Zhao on 2021-04-09 06:04:30 CEST --- # oc version Client Version: 4.8.0-0.nightly-2021-04-08-200632 Server Version: 4.8.0-0.nightly-2021-04-08-200632 Kubernetes Version: v1.21.0-rc.0+6d27558 checked from prometheus, query parameter: count(apiserver_requested_deprecated_apis{removed_release="1.22"}) by(instance,version,group,resource) version is v1beta1 {group="certificates.k8s.io", instance="10.0.160.188:6443", resource="certificatesigningrequests", version="v1beta1"} 1 {group="extensions", instance="10.0.160.188:6443", resource="ingresses", version="v1beta1"} 1 {group="rbac.authorization.k8s.io", instance="10.0.160.188:6443", resource="clusterrolebindings", version="v1beta1"} 1 {group="rbac.authorization.k8s.io", instance="10.0.160.188:6443", resource="rolebindings", version="v1beta1"} 1 {group="rbac.authorization.k8s.io", instance="10.0.160.188:6443", resource="roles", version="v1beta1"} 1 {group="admissionregistration.k8s.io", instance="10.0.160.188:6443", resource="mutatingwebhookconfigurations", version="v1beta1"} 1 {group="admissionregistration.k8s.io", instance="10.0.160.188:6443", resource="validatingwebhookconfigurations", version="v1beta1"} 1 {group="apiextensions.k8s.io", instance="10.0.160.188:6443", resource="customresourcedefinitions", version="v1beta1"} 1 but the api versions are all actually v1, which means apiserver_requested_deprecated_apis may post the wrong result # for i in certificatesigningrequests ingresses clusterrolebindings rolebindings roles mutatingwebhookconfigurations validatingwebhookconfigurations customresourcedefinitions; do oc api-resources | grep $i; echo -e "\n"; done certificatesigningrequests csr certificates.k8s.io/v1 false CertificateSigningRequest ingresses config.openshift.io/v1 false Ingress ingresses ing extensions/v1beta1 true Ingress ingresses ing networking.k8s.io/v1 true Ingress clusterrolebindings authorization.openshift.io/v1 false ClusterRoleBinding clusterrolebindings rbac.authorization.k8s.io/v1 false ClusterRoleBinding clusterrolebindings authorization.openshift.io/v1 false ClusterRoleBinding rolebindings authorization.openshift.io/v1 true RoleBinding clusterrolebindings rbac.authorization.k8s.io/v1 false ClusterRoleBinding rolebindings rbac.authorization.k8s.io/v1 true RoleBinding clusterroles authorization.openshift.io/v1 false ClusterRole roles authorization.openshift.io/v1 true Role clusterroles rbac.authorization.k8s.io/v1 false ClusterRole roles rbac.authorization.k8s.io/v1 true Role mutatingwebhookconfigurations admissionregistration.k8s.io/v1 false MutatingWebhookConfiguration validatingwebhookconfigurations admissionregistration.k8s.io/v1 false ValidatingWebhookConfiguration customresourcedefinitions crd,crds apiextensions.k8s.io/v1 false CustomResourceDefinition --- Additional comment from Stefan Schimanski on 2021-05-18 16:06:30 CEST --- --- Additional comment from Martin Bukatovic on 2021-05-19 17:18:07 CEST --- The alert this bug talks about is APIRemovedInNextEUSReleaseInUse. I'm mentioning it here so that it's possible to find this bug when one searches by content of bugzilla comments.
In QE CI env that runs many cases (created many projects for the cases), checked not only KCM, but also below kube-system:namespace-controller: $ MASTERS=`oc get no | grep master | grep -o '^[^ ]*'` $ for i in $MASTERS; do oc debug no/$i -- chroot /host bash -c "grep -hE '"'"k8s.io/removed-release":"[^"]+"'"' /var/log/kube-apiserver/audit*.log" ; done > all.log $ grep '"k8s.io/removed-release":"1.22"' all.log > 1.22.log $ jq -r '.user.username+": "+.requestURI' 1.22.log | sed 's/=[0-9][^&]*/=***/g' | sort | uniq -c | sort -n > 1.22.removed.apis $ cat 1.22.removed.apis | grep -v kube-controller-manager 2 system:serviceaccount:kube-system:namespace-controller: /apis/extensions/v1beta1/namespaces/01ptm/ingresses 2 system:serviceaccount:kube-system:namespace-controller: /apis/extensions/v1beta1/namespaces/05mpl/ingresses 2 system:serviceaccount:kube-system:namespace-controller: /apis/extensions/v1beta1/namespaces/0hqei/ingresses 2 system:serviceaccount:kube-system:namespace-controller: /apis/extensions/v1beta1/namespaces/0t7dk/ingresses ...snipped, totally 352 lines...
Checked the PR code and verified in 4.8.0-0.nightly-2021-06-14-145150: Positive testing: Checking env, comment 1 still exists, and kube-controller-manager still accesses ingresses.v1beta1.extensions. Given this, the Alerting page does not show APIRemovedInNextReleaseInUse alert in Firing or Pending state. This means the alert already excludes KCM. Negative testing: OAS_SA_TOKEN=`oc sa get-token openshift-apiserver-sa -n openshift-apiserver` oc login --token "$OAS_SA_TOKEN" for i in {1..100}; do oc get ingresses.v1beta1.extensions; done And checking metrics: sum by(system_client) (rate(apiserver_request_total{resource="ingresses",version="v1beta1"}[4h])) , there is one item with empty system_client. Given this, the Alerting page now shows APIRemovedInNextReleaseInUse alert in Firing or Pending state. This means the alert still works.
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 (Moderate: OpenShift Container Platform 4.8.2 bug fix and security update), 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/RHSA-2021:2438