Whizard 组件资源配额推荐

默认各组件资源配置

工作负载名称工作负载类型副本数容器名称CPU 限制内存限制CPU 请求内存请求说明
gateway-whizardDeployment2gateway24Gi50m64Mi
ingester-whizardStatefulSet2receive416Gi50m64Mi为软租户单独配置
ingester-whizard-(local/remote)-auto-0StatefulSet2receive416Gi50m64Mi将根据租户数量自动扩展
compactor-whizard-remote-xxxStatefulSet1receive416Gi50m64Mi将根据租户数量自动扩展
store-whizard-remote-xxxStatefulSet2receive416Gi50m64Mi将根据租户数量自动扩展
query-frontend-whizardDeployment2query-frontend24Gi50m64Mi
query-whizardDeployment3query24Gi50m64Mi
proxy24Gi50m64Misidecar 容器
router-whizardDeployment2receive24Gi50m64Mi
ruler-<tenant>-0StatefulSet1rule24Gi50m64Mi与 Whizard Tenant 一一对应
write-proxy24Gi50m64Misidecar 容器
config-reloader100m128Mi100m128Misidecar 容器
whizard-adapterDeployment1adapter200m256Mi100m128Mi
whizard-agent-proxyDeployment1whizard-agent-proxy24Gi50m64Mi
whizard-controller-managerDeployment1manager200m512Mi100m128Mi

资源使用配置调整策略

  • Query: 与数据查询压力相关,在长时间范围查询或查询结果数据量大时负载较高,正常情况下使用默认配置即可。如需调整,每次增大 50%
  • Ingester: Ingester 的资源消耗与指标摄入率(即平均每秒采集Series 数量)相关,根据压测建议每增加10k serise 增加1C 4G
  • Compactor: Compactor 需要拉取数据到本地进行压缩与降采样,资源调整建议每次增加20%,并注意挂载卷使用情况
  • Store: 当对应存储数据保存较久,建议使用时间分片进行拓展,参考 Store 基于时间分片进行扩展
  • 其他组件遵循常见配额调整策略,如需调整,每次增大 20%