写在前面

controller-manager 是 Kubernetes 控制面的组件，通常不太可能出问题，一般监控一下通用的进程指标就问题不大了，不过 controller-manager 确实也暴露了很多​​/metrics​​白盒指标，我们也一并梳理一下相关内容。

黑盒测试

类似上一篇《​​Kubernetes监控手册06-监控APIServer​​》描述的方法，我们先从黑盒角度测试一下，看看 controller-manager 的​​/metrics​​接口是否直接可用。

[root@tt-fc-dev01.nj manifests]# ss -tlnp|grep controllerLISTEN 0      128                *:10257            *:*    users:(("kube-controller",pid=2782446,fd=7))[root@tt-fc-dev01.nj manifests]# curl -s http://localhost:10257/metricsClient sent an HTTP request to an HTTPS server.[root@tt-fc-dev01.nj manifests]# curl -k -s https://localhost:10257/metrics{  "kind": "Status",  "apiVersion": "v1",  "metadata": {},  "status": "Failure",  "message": "forbidden: User \"system:anonymous\" cannot get path \"/metrics\"",  "reason": "Forbidden",  "details": {},  "code": 403}

看起来也是需要认证的，我们直接复用上一篇创建的 Token，看看能否拿到数据：

(资料图片)

[root@tt-fc-dev01.nj yamls]# token=`kubectl get secret categraf-token-6whbs -n flashcat -o jsonpath={.data.token} | base64 -d`[root@tt-fc-dev01.nj yamls]# curl -s -k -H "Authorization: Bearer $token" https://localhost:10257/metrics > cm.metrics[root@tt-fc-dev01.nj yamls]# head -n 6 cm.metrics# HELP apiserver_audit_event_total [ALPHA] Counter of audit events generated and sent to the audit backend.# TYPE apiserver_audit_event_total counterapiserver_audit_event_total 0# HELP apiserver_audit_requests_rejected_total [ALPHA] Counter of apiserver requests rejected due to an error in audit logging backend.# TYPE apiserver_audit_requests_rejected_total counterapiserver_audit_requests_rejected_total 0[root@tt-fc-dev01.nj yamls]# cat cm.metrics | wc -l10070

妥了，可以复用之前的 Token。

配置采集

我们还是使用 Prometheus agent mode 来拉取数据，原汁原味的，只要把 controller-manager 部分也加上就行了。改造之后的 prometheus-agent-configmap.yaml 内容如下：

apiVersion: v1kind: ConfigMapmetadata:  name: prometheus-agent-conf  labels:    name: prometheus-agent-conf  namespace: flashcatdata:  prometheus.yml: |-    global:      scrape_interval: 15s      evaluation_interval: 15s    scrape_configs:      - job_name: "apiserver"        kubernetes_sd_configs:        - role: endpoints        scheme: https        tls_config:          insecure_skip_verify: true        authorization:          credentials_file: /var/run/secrets/kubernetes.io/serviceaccount/token        relabel_configs:        - source_labels: [__meta_kubernetes_namespace, __meta_kubernetes_service_name, __meta_kubernetes_endpoint_port_name]          action: keep          regex: default;kubernetes;https      - job_name: "controller-manager"        kubernetes_sd_configs:        - role: endpoints        scheme: https        tls_config:          insecure_skip_verify: true        authorization:          credentials_file: /var/run/secrets/kubernetes.io/serviceaccount/token        relabel_configs:        - source_labels: [__meta_kubernetes_namespace, __meta_kubernetes_service_name, __meta_kubernetes_endpoint_port_name]          action: keep          regex: kube-system;kube-controller-manager;https    remote_write:    - url: "http://10.206.0.16:19000/prometheus/v1/write"

这里我新增了一个scrape job name：controller-manager，Kubernetes 服务发现仍然使用 endpoints，匹配规则有三点(通过 relabel_configs 的 keep 实现)：

​​__meta_kubernetes_namespace​​endpoint 的 namespace 要求是 kube-system​​__meta_kubernetes_service_name​​service name 要求是 kube-controller-manager​​__meta_kubernetes_endpoint_port_name​​endpoint 的 port_name 要求是叫 https

如果你没有采集成功，就要去看看有没有这个 endpoint：

[work@tt-fc-dev01.nj yamls]$ kubectl get endpoints -n kube-systemNAME                      ENDPOINTS                                                            AGEetcd                      10.206.0.16:2381                                                     126detcd-service              10.206.0.16:2379                                                     75detcd-service2             10.206.10.16:2379                                                    75dkube-controller-manager   10.206.0.16:10257                                                    74dkube-dns                  172.16.0.85:53,172.16.1.4:53,172.16.0.85:53 + 3 more...              324dkube-scheduler            10.206.0.16:10259                                                    131dkube-state-metrics        172.16.3.198:8081,172.16.3.198:8080                                  75dkubelet                   10.206.0.11:10250,10.206.0.16:10250,10.206.0.17:10250 + 15 more...   315d[work@tt-fc-dev01.nj yamls]$ kubectl get endpoints -n kube-system kube-controller-manager -o yamlapiVersion: v1kind: Endpointsmetadata:  annotations:    endpoints.kubernetes.io/last-change-trigger-time: "2022-09-15T09:43:21Z"  creationTimestamp: "2022-09-15T09:43:21Z"  labels:    k8s-app: kube-controller-manager  name: kube-controller-manager  namespace: kube-system  resourceVersion: "112212043"  uid: 52cfb383-6d2b-452e-9a1f-95c7a898a1b4subsets:- addresses:  - ip: 10.206.0.16    nodeName: 10.206.0.16    targetRef:      kind: Pod      name: kube-controller-manager-10.206.0.16      namespace: kube-system      resourceVersion: "112211925"      uid: d9515495-057c-4ea6-ad1f-28341498710f  ports:  - name: https    port: 10257    protocol: TCP

​​__meta_kubernetes_endpoint_port_name​​就是上面的倒数第三行。这些信息我的环境里都是有的，如果你的环境没有对应的 endpoint，可以手工创建一个 service，孔飞老师之前给大家准备过一个​​https://github.com/flashcatcloud/categraf/blob/main/k8s/controller-service.yaml​​，把这个 controller-service.yaml apply 一下就行了。另外，如果是用 kubeadm 安装的 controller-manager，还要记得修改​​/etc/kubernetes/manifests/kube-controller-manager.yaml​​，调整 controller-manager 的启动参数：​​--bind-address=0.0.0.0​​。

监控大盘

controller-manager 的大盘已经准备好了，地址在​​https://github.com/flashcatcloud/categraf/blob/main/k8s/cm-dash.json​​，可以直接导入夜莺使用。如果觉得大盘有需要改进的地方，欢迎PR。

监控指标

controller-manager 的关键指标分别是啥意思，孔飞老师之前整理过，我给搬过来了：

# HELP rest_client_request_duration_seconds [ALPHA] Request latency in seconds. Broken down by verb and URL.# TYPE rest_client_request_duration_seconds histogram请求apiserver的耗时分布，按照url+verb统计# HELP cronjob_controller_cronjob_job_creation_skew_duration_seconds [ALPHA] Time between when a cronjob is scheduled to be run, and when the corresponding job is created# TYPE cronjob_controller_cronjob_job_creation_skew_duration_seconds histogramcronjob 创建到运行的时间分布# HELP leader_election_master_status [ALPHA] Gauge of if the reporting system is master of the relevant lease, 0 indicates backup, 1 indicates master. "name" is the string used to identify the lease. Please make sure to group by name.# TYPE leader_election_master_status gauge控制器的选举状态，0表示backup， 1表示master # HELP node_collector_zone_health [ALPHA] Gauge measuring percentage of healthy nodes per zone.# TYPE node_collector_zone_health gauge每个zone的健康node占比# HELP node_collector_zone_size [ALPHA] Gauge measuring number of registered Nodes per zones.# TYPE node_collector_zone_size gauge每个zone的node数# HELP process_cpu_seconds_total Total user and system CPU time spent in seconds.# TYPE process_cpu_seconds_total countercpu使用量（也可以理解为cpu使用率）# HELP process_open_fds Number of open file descriptors.# TYPE process_open_fds gauge控制器打开的fd数# HELP pv_collector_bound_pv_count [ALPHA] Gauge measuring number of persistent volume currently bound# TYPE pv_collector_bound_pv_count gauge当前绑定的pv数量# HELP pv_collector_unbound_pvc_count [ALPHA] Gauge measuring number of persistent volume claim currently unbound# TYPE pv_collector_unbound_pvc_count gauge当前没有绑定的pvc数量 # HELP pv_collector_bound_pvc_count [ALPHA] Gauge measuring number of persistent volume claim currently bound# TYPE pv_collector_bound_pvc_count gauge当前绑定的pvc数量# HELP pv_collector_total_pv_count [ALPHA] Gauge measuring total number of persistent volumes# TYPE pv_collector_total_pv_count gaugepv总数量# HELP workqueue_adds_total [ALPHA] Total number of adds handled by workqueue# TYPE workqueue_adds_total counter各个controller已接受的任务总数与apiserver的workqueue_adds_total指标类似# HELP workqueue_depth [ALPHA] Current depth of workqueue# TYPE workqueue_depth gauge各个controller队列深度，表示一个controller中的任务的数量与apiserver的workqueue_depth类似，这个是指各个controller中队列的深度，数值越小越好# HELP workqueue_queue_duration_seconds [ALPHA] How long in seconds an item stays in workqueue before being requested.# TYPE workqueue_queue_duration_seconds histogram任务在队列中的等待耗时,按照控制器分别统计# HELP workqueue_work_duration_seconds [ALPHA] How long in seconds processing an item from workqueue takes.# TYPE workqueue_work_duration_seconds histogram任务出队到被处理完成的时间，按照控制分别统计# HELP workqueue_retries_total [ALPHA] Total number of retries handled by workqueue# TYPE workqueue_retries_total counter任务进入队列重试的次数# HELP workqueue_longest_running_processor_seconds [ALPHA] How many seconds has the longest running processor for workqueue been running.# TYPE workqueue_longest_running_processor_seconds gauge正在处理的任务中，最长耗时任务的处理时间# HELP endpoint_slice_controller_syncs [ALPHA] Number of EndpointSlice syncs# TYPE endpoint_slice_controller_syncs counterendpoint_slice 同步的数量(1.20以上)# HELP get_token_fail_count [ALPHA] Counter of failed Token() requests to the alternate token source# TYPE get_token_fail_count counter获取token失败的次数# HELP go_memstats_gc_cpu_fraction The fraction of this program"s available CPU time used by the GC since the program started.# TYPE go_memstats_gc_cpu_fraction gaugecontroller gc的cpu使用率

相关文章

​​Kubernetes监控手册01-体系介绍​​​​Kubernetes监控手册02-宿主监控概述​​​​Kubernetes监控手册03-宿主监控实操​​​​Kubernetes监控手册04-监控Kube-Proxy​​​​Kubernetes监控手册05-监控Kubelet​​​​Kubernetes监控手册06-监控APIServer​​

关于作者

本文作者秦晓辉，​​Flashcat​​合伙人，文章内容是Flashcat技术团队共同沉淀的结晶，作者做了编辑整理，我们会持续输出监控、稳定性保障相关的技术文章，文章可转载，转载请注明出处，尊重技术人员的成果。

如果对 Nightingale、Categraf、Prometheus 等技术感兴趣，欢迎加入我们的微信群组，联系我(picobyte)拉入部落，和社区同仁一起探讨监控技术。