n9e categraf k8s监控配置 -cadvisor

1、创建用户角色权限

bash 复制代码
vi  z-categraf-monitor-rbac.yaml
bash 复制代码
apiVersion: v1
kind: ServiceAccount
metadata:
  name: z-categraf-monitor
  namespace: z-monitor
  labels:
    app: categraf
    purpose: cadvisor-monitor
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: z-categraf-monitor-clusterrole
  labels:
    app: categraf
    purpose: cadvisor-monitor
rules:
  - nonResourceURLs:
      - "/metrics/cadvisor"
      - "/stats/summary"
      - "/pods"
      - "/nodes/proxy/metrics/cadvisor"
    verbs: ["get"]
  - apiGroups: [""]
    resources:
      - "nodes"
      - "nodes/metrics"
      - "nodes/stats"
      - "nodes/proxy"
    verbs: ["get", "list", "watch"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: z-categraf-monitor-clusterrolebinding
  labels:
    app: categraf
    purpose: cadvisor-monitor
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: z-categraf-monitor-clusterrole
subjects:
  - kind: ServiceAccount
    name: z-categraf-monitor
    namespace: z-monitor
bash 复制代码
kubectl apply -f z-categraf-monitor-rbac.yaml

2、获取token

bash 复制代码
# 提取Token
SECRET_NAME=$(kubectl get sa z-categraf-monitor -n z-monitor -o jsonpath='{.secrets[0].name}')
VALID_TOKEN=$(kubectl get secret $SECRET_NAME -n z-monitor -o jsonpath='{.data.token}' | base64 -d)

#  将Token写入文件
echo "$VALID_TOKEN" > /data/zz/z-categraf-monitor.token

3、每个节点配置 categraf/conf/input.cadvisor/cim-pord-cadvisor.toml

bash 复制代码
vi  cim-pord-cadvisor.toml
bash 复制代码
# # collect interval
interval = 15

[[instances]]
url = "https://127.0.0.1:10250"
type = "kubelet"

bearer_token_string = "eyxxxxx"
ignore_label_keys = ["id","name", "container_label*"]
insecure_skip_verify = true
use_tls = true
container_exclude = ["^POD$"]
# ignore_label_keys = ["id","name", "container_label*"]
## choose_label_keys = ["id"]

timeout = "15s"# # collect interval
 
相关推荐
weirong1 分钟前
Operator 的测试该写到哪一层?139 个用例之后我划的三条线
kubernetes
阿里云云原生30 分钟前
稳定复现超 70% 的缺陷,报警到自动修复的 Agent 流水线实践
云原生
阿里云云原生1 小时前
从“自证清白”到“因果链条”:智能体如何重塑运维排障效率
云原生
阿里云云原生2 小时前
TechTalk丨从能买奶茶到能进企业,Agent 绕不开同一组问题
云原生·agent
逐流人3 小时前
手工建集群到全自动发布:k8s集群部署与DevOps流水线整合实战
ci/cd·云原生·kubernetes·云计算·devops
我不会起名字3224 小时前
K8s Pod 反复重启 CrashLoopBackOff:从 exit code 到根因的 5 步排查
云原生·容器·kubernetes·k8s
瑞克-桑切斯4 小时前
ERP 企业生产管理系统部署手册
运维·docker·云原生·容器·eureka
我不会起名字3224 小时前
K8s 节点磁盘打满导致 Pod 被驱逐:4 类占用与回收顺序
运维·kubernetes·k8s·diskpressure·evicted
孔令飞5 小时前
把 Agent 搬进飞书:飞书 Agent 机器人演示
云原生·go·ai编程
瑞克-桑切斯5 小时前
docker mail商城
运维·docker·容器