告警规则
常用告警规则配置
-
alerts
## CPU告警规则 groups: - name: CpuAlertRule rules: - alert: PodCPU告警 expr: onecore:pod > 80 or twocore:pod / 2 > 80 or squarecore:pod / 4 > 80 for: 2m labels: severity: warning annotations: description: "CPU使用率大于80%" value: "{{$value}}%" #summary: 'CPU使用率大于80%，当前值为{{.Value}}%，CPU使用率: {{ printf `ceil(100 - ((avg by (instance)(irate(node_cpu_seconds_total{mode="idle",instance="%s"}[1m]))) *100))` $labels.instance | query | first | value }}%' - alert: NodeCPU告警 expr: round(100-avg(irate(node_cpu_seconds_total{mode="idle"}[5m]))by(kubernetes_node)*100) > 80 for: 2m labels: severity: warning annotations: description: "CPU使用率大于80%" value: "{{$value}}%" #summary: 'CPU使用率大于80%，当前值为{{.Value}}%，CPU使用率: {{ printf `ceil(100 - ((avg by (instance)(irate(node_cpu_seconds_total{mode="idle",instance="%s"}[1m]))) *100))` $labels.instance | query | first | value }}%' ## DISK告警规则 - name: DiskAlertRule rules: - alert: Pod磁盘告警 expr: round(container_fs_usage_bytes{container=~".+",container!~"POD"}/1024/1024/1024/10*100) > 85 for: 1m labels: severity: warning annotations: description: "磁盘使用率大于85%" value: "{{$value}}%" - alert: Node磁盘告警 expr: round((1- node_filesystem_avail_bytes{fstype=~"ext.+|nfs.+",mountpoint!~".*docker.*"}/node_filesystem_size_bytes{fstype=~"ext.+|nfs.+",mountpoint!~".*docker.*"})*100) > 85 for: 1m labels: severity: warning annotations: description: "磁盘使用率大于85%" value: "{{$value}}%" ## MEM告警规则 - name: MemAlertRule rules: - alert: Pod内存告警 expr: round(container_memory_usage_bytes{container=~".+",container!~"POD|.+reload",pod!~"^csi.+"}/container_spec_memory_limit_bytes{container=~".+",container!~"POD|.reload",pod!~"^csi.+"}*100) > 85 for: 2m labels: severity: warning annotations: description: "内存使用率大于85%" value: "{{$value}}%" - alert: Node内存告警 expr: round(100-((node_memory_MemAvailable_bytes*100)/node_memory_MemTotal_bytes)) > 80 for: 2m labels: severity: warning annotations: description: "内存使用率大于85%" value: "{{$value}}%" ## Pod意外重启 - name: PodRestartAlertRule rules: - alert: Pod重启告警 expr: delta(kube_pod_container_status_restarts_total[1m]) > 0 for: 1s labels: severity: warning annotations: description: "Pod发生意外重启事件" ## JvmCMSOldGC - name: PodJvmOldGCAlertRule rules: - alert: PodJvmCMSOldGC expr: round((jvm_memory_pool_bytes_used{pool=~".+Old Gen"}/jvm_memory_pool_bytes_max{pool=~".+Old Gen"})*100) > 89 for: 5s labels: severity: warning annotations: description: "Pod堆内存触发CMSOldGC" value: "{{$value}}%" ## Pod实例异常 - name: ContainerInstanceAlertRule rules: - alert: Pod实例异常 expr: kube_pod_container_status_ready - kube_pod_container_status_running > 0 for: 20s labels: severity: warning annotations: description: "Container实例异常" ## Pod实例OOM - name: ContainerOOMAlertRule rules: - alert: Pod实例OOM expr: kube_pod_container_status_terminated_reason{reason="OOMKilled"} > 0 for: 1s labels: severity: warning annotations: description: "Container实例OOM" ## Pod实例驱逐 - name: ContainerEvictionAlertRule rules: - alert: Pod实例驱逐 expr: kube_pod_container_status_terminated_reason{reason="Evicted"} > 0 for: 1s labels: severity: warning annotations: description: "Container实例驱逐" ## MQ内存告警 - name: MQMemoryAlertRule rules: - alert: MQ内存水位线 expr: rabbitmq_node_mem_alarm{job=~".*rabbitmq.*"} == 1 for: 1s labels: severity: warning annotations: description: "RabbitMQ内存高水位线告警" summary: RabbitMQ {{`{{ $labels.instance }}`}} High Memory Alarm is going off. Which means the node hit highwater mark and has cut off network connectivity, see RabbitMQ WebUI - alert: MQ内存使用告警 expr: round(avg(rabbitmq_node_mem_used{job=~".*rabbitmq.*"} / rabbitmq_node_mem_limit{job=~".*rabbitmq.*"})by(node,kubernetes_namespace)*100) > 90 for: 10s labels: severity: warning annotations: description: "RabbitMQ使用告警" value: "{{$value}}%" summary: RabbitMQ {{`{{ $labels.instance }}`}} Memory Usage > 90% ##PodJava进程异常 - name: PodJavaProcessAlertRule rules: - alert: PodJava进程异常 expr: sum(up{job="kubernetes-pods-jvm"})by(kubernetes_container_name,kubernetes_pod_name) == 0 for: 10s labels: severity: warning annotations: description: "PodJava进程异常" summary: "赶快看看吧，顶不住了"
-
recording_rules
groups: - name: CpuRecordRules rules: - record: onecore:pod expr: round(sum by(pod, container, instance, namespace, name) (irate(container_cpu_usage_seconds_total{container!~"|POD|prod-xianxiang-edu-loan|prod-xy-fund|prod-common-callcenter|prod-risk-service|prod-qn-web-api|prod-xc-fund|prod-xc-user|sys-ingress|etcd|prod-qn-mp|prod-xc-common|prod-xianxiang-zuul|prod-qn-user|prod-xc-riskapi|prod-common-message|prod-common-trust-service|prod-xc-collection|kube-controller-manager|prod-qn-risk|prod-xy-zuul|metrics-server|prod-nflow-manager|kube-scheduler|prod-qn-gateway|prod-xc-pay|coredns|kube-apiserver|prod-qn-oms|prod-common-service|prod-nfsp-service|pord-ingress|prod-qn-cms|prod-internal-ingress|prod-xc-loan|prod-rabbitmq|prometheus-server"}[5m]) * 100)) - record: twocore:pod expr: round(sum by(pod, container, instance, namespace, name) (irate(container_cpu_usage_seconds_total{container=~"prod-xianxiang-edu-loan|prod-xy-fund|prod-common-callcenter|prod-risk-service|prod-qn-web-api|prod-xc-fund|prod-xc-user|sys-ingress|etcd|prod-qn-mp|prod-xc-common|prod-xianxiang-zuul|prod-qn-user|prod-xc-riskapi|prod-common-message|prod-common-trust-service|prod-xc-collection|kube-controller-manager|prod-qn-risk|prod-xy-zuul|metrics-server|prod-nflow-manager|kube-scheduler|prod-qn-gateway|prod-xc-pay|coredns|kube-apiserver|prod-qn-oms|prod-common-service|prod-nfsp-service|pord-ingress|prod-qn-cms|prod-internal-ingress|prod-xc-loan"}[5m]) * 100)) - record: squarecore:pod expr: round(sum by(pod, container, instance, namespace, name) (irate(container_cpu_usage_seconds_total{container=~"prod-rabbitmq|prometheus-server"}[5m]) * 100))