Kubernetes API Server¶
Kubernetes API Server 性能指标展示,包括请求数、工作队列增速、工作队列深度、 CPU 、 Memory 、 Goroutine 等
配置¶
版本支持¶
-
操作系统支持:Linux
-
Kubernetes 版本:1.18+
前置条件¶
-
已部署 DataKit,请参考 Kubernetes 集群 <安装 Datakit>
-
采集 Kubernetes API Server 指标数据,需要 Kubernetes 安装 Metrics-Server 组件。
指标采集¶
- 使用
yaml
创建bearer-token
授权信息
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: bearer-token
rules:
- apiGroups: [""]
resources:
- nodes
- nodes/metrics
- nodes/stats
- nodes/proxy
- services
- endpoints
- pods
- configmaps
- secrets
- resourcequotas
- replicationcontrollers
- limitranges
- persistentvolumeclaims
- persistentvolumes
- namespaces
verbs: ["get", "list", "watch"]
- apiGroups:
- extensions
- networking.k8s.io
resources:
- ingresses
verbs: ["get", "list", "watch"]
- nonResourceURLs: ["/metrics", "/metrics/cadvisor"]
verbs: ["get"]
---
apiVersion: v1
kind: ServiceAccount
metadata:
name: bearer-token
namespace: default
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
name: bearer-token
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: ClusterRole
name: bearer-token
subjects:
- kind: ServiceAccount
name: bearer-token
namespace: default
- 获取bearer_token
kubectl get secret `kubectl get secret -ndefault | grep bearer-token | awk '{print $1}'` -o jsonpath={.data.token} | base64 -d
- ConfigMap 增加
api-server.conf
配置
在部署 DataKit 使用的 datakit.yaml
文件中,ConfigMap 资源中增加 api-server.conf
。
apiVersion: v1
kind: ConfigMap
metadata:
name: datakit-conf
namespace: datakit
data:
#### api-server ##下面是新增部分
api-server.conf: |-
[[inputs.prom]]
## Exporter地址或者文件路径(Exporter地址要加上网络协议http或者https)
## 文件路径各个操作系统下不同
## Windows example: C:\\Users
## UNIX-like example: /usr/local/
url = "https://172.31.16.148:6443/metrics"
## 采集器别名
source = "prom-api-server"
## 指标类型过滤, 可选值为 counter, gauge, histogram, summary
# 默认只采集 counter 和 gauge 类型的指标
# 如果为空,则不进行过滤
metric_types = ["counter", "gauge"]
## 指标名称过滤
# 支持正则,可以配置多个,即满足其中之一即可
# 如果为空,则不进行过滤
#metric_name_filter = [""]
## 指标集名称前缀
# 配置此项,可以给指标集名称添加前缀
#measurement_prefix = "prom_api_server"
## 指标集名称
# 默认会将指标名称以下划线"_"进行切割,切割后的第一个字段作为指标集名称,剩下字段作为当前指标名称
# 如果配置measurement_name, 则不进行指标名称的切割
# 最终的指标集名称会添加上measurement_prefix前缀
measurement_name = "prom_api_server"
## 采集间隔 "ns", "us" (or "µs"), "ms", "s", "m", "h"
interval = "60s"
## 过滤tags, 可配置多个tag
# 匹配的tag将被忽略
tags_ignore = ["apiservice","bound","build_date","compiler","component","crd","dry_run","endpoint","error_type","flow_schema","git_commit","git_tree_state","git_version","go_version","group","grpc_code","grpc_method","grpc_service","grpc_type","kind","major","method","minor","operation","platform","priority_level","reason","rejection_code","removed_release","request_kind","resource","result","scope","source","status","subresource","type","usage","username","verb","version"]
metric_name_filter = ["workqueue_adds_total","workqueue_depth","apiserver_request_total","process_resident_memory_bytes","process_cpu_seconds_total","go_goroutines"]
## TLS 配置
tls_open = true
tls_ca = "/var/run/secrets/kubernetes.io/serviceaccount/ca.crt"
#tls_cert = "/etc/kubernetes/pki/apiserver.crt"
#tls_key = "/etc/kubernetes/pki/apiserver.key"
## 自定义指标集名称
# 可以将包含前缀prefix的指标归为一类指标集
# 自定义指标集名称配置优先measurement_name配置项
#[[inputs.prom.measurements]]
# prefix = ""
# name = ""
## 自定义认证方式,目前仅支持 Bearer Token
[inputs.prom.auth]
type = "bearer_token"
token = "eyJhbGciOiJSUzI1NiIsImtpZCI6ImlfX2V6UXdXWkpKUWZ6QlBxTGdSRTBpa0J1a2VpQUU3Q0JMWGFfYWNDYWcifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJkZWZhdWx0Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZWNyZXQubmFtZSI6ImJlYXJlci10b2tlbi10b2tlbi05emI5dCIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50Lm5hbWUiOiJiZWFyZXItdG9rZW4iLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlcnZpY2UtYWNjb3VudC51aWQiOiJkNWQxNDkzNi00NmM1LTRjZjMtYmI2MS00ODhhOTFiYTRjMTQiLCJzdWIiOiJzeXN0ZW06c2VydmljZWFjY291bnQ6ZGVmYXVsdDpiZWFyZXItdG9rZW4ifQ.sBQUGE67N6BV6mnC0g72k8ciiSjEZ-ctFjHcyiP_rBp9paUnGwd3ouheF0ddGormn6esOGR1t6vvDdta9BiE3i5mHpJsOifkVXzv85N3qllJfSpXvIIn-LNq-wxnK55QbOhXQjeFKF0PBanJk4m_kWCM6SOuFrH9s8cHGhKEVCYw_7ScUwHCDGQVUq_zKCfKll20GHSwhlzjjt2tz07UYdQs5kQ9AN8VbM9qNIJmpasPOeqod9hTbevnL3kO5Lcd4h4NUOT8JfJ2Om72NvH71-xWNH0U_Hqf2yS0_ZlnneBESq4FDjbm1VnJPxeIOJL0dMaoRJVPPtA0yUhX5MYV7A"
#token_file = "/var/run/secrets/kubernetes.io/serviceaccount/token"
## 自定义Tags
[inputs.prom.tags]
instance = "172.31.16.148:6443"
-
参数说明:
-
url:api-server metrics 地址
- source:采集器别名
- metric_types:指标类型过滤
- metric_name_filter:指标名称过滤
- measurement_prefix:指标集名称前缀
- measurement_name:指标集名称
- interval:采集间隔
- tags_ignore: 忽略的 tag
- metric_name_filter: 保留的指标名
- tls_open:是否忽略安全验证 (如果是 HTTPS,请设置为 true,并设置相应证书),此处为 true
- tls_ca:ca 证书路径
- type:自定义认证方式,api-server 使用 bearer_token 认证
- token_file:认证文件路径
-
inputs.prom.tags:请参考插件标签
-
挂载
api-server.conf
在 datakit.yaml
文件的 volumeMounts
下面增加下面内容。
- mountPath: /usr/local/datakit/conf.d/prom/api-server.conf
name: datakit-conf
subPath: api-server.conf
- Kubernetes默认已暴露指标,可以直接通过 curl 方式来查看相关指标。
curl -k "https://172.31.16.148:6443/metrics" -H "Authorization: Bearer eyJhbGciOiJSUzI1NiIsImtpZCI6ImlfX2V6UXdXWkpKUWZ6QlBxTGdSRTBpa0J1a2VpQUU3Q0JMWGFfYWNDYWcifQ.eyJpc3MiOiJrdWJlcm5ldGVzL3NlcnZpY2VhY2NvdW50Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9uYW1lc3BhY2UiOiJkZWZhdWx0Iiwia3ViZXJuZXRlcy5pby9zZXJ2aWNlYWNjb3VudC9zZWNyZXQubmFtZSI6ImJlYXJlci10b2tlbi10b2tlbi05emI5dCIsImt1YmVybmV0ZXMuaW8vc2VydmljZWFjY291bnQvc2VydmljZS1hY2NvdW50Lm5hbWUiOiJiZWFyZXItdG9rZW4iLCJrdWJlcm5ldGVzLmlvL3NlcnZpY2VhY2NvdW50L3NlcnZpY2UtYWNjb3VudC51aWQiOiJkNWQxNDkzNi00NmM1LTRjZjMtYmI2MS00ODhhOTFiYTRjMTQiLCJzdWIiOiJzeXN0ZW06c2VydmljZWFjY291bnQ6ZGVmYXVsdDpiZWFyZXItdG9rZW4ifQ.sBQUGE67N6BV6mnC0g72k8ciiSjEZ-ctFjHcyiP_rBp9paUnGwd3ouheF0ddGormn6esOGR1t6vvDdta9BiE3i5mHpJsOifkVXzv85N3qllJfSpXvIIn-LNq-wxnK55QbOhXQjeFKF0PBanJk4m_kWCM6SOuFrH9s8cHGhKEVCYw_7ScUwHCDGQVUq_zKCfKll20GHSwhlzjjt2tz07UYdQs5kQ9AN8VbM9qNIJmpasPOeqod9hTbevnL3kO5Lcd4h4NUOT8JfJ2Om72NvH71-xWNH0U_Hqf2yS0_ZlnneBESq4FDjbm1VnJPxeIOJL0dMaoRJVPPtA0yUhX5MYV7A"
- 重启 DataKit
插件标签 (必选)¶
参数说明
- 该配置为自定义标签,可以填写任意 key-value 值
- 以下示例配置完成后,api-server 指标都会带有如下类似的标签,可以进行快速查询
- 采集 api-server 指标,必填的 key 是 instance,值是 api-server 的地址
重启 DataKit
指标¶
指标 | 描述 | 数据类型 |
---|---|---|
apiserver_request_total |
Counter of apiserver requests broken out for each verb, dry run value, group, version, resource, scope, component, and HTTP response code. | int |
workqueue_adds_total |
Total number of adds handled by workqueue | int |
workqueue_depth |
Current depth of workqueue | int |
process_resident_memory_bytes |
Resident memory size in bytes | B |
process_cpu_seconds_total |
Total user and system CPU time spent in seconds | float |
go_goroutines |
Number of goroutine that currently exist | int |
常见问题排查¶
<无数据上报排查>