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-whizard	Deployment	2	gateway	2	4Gi	50m	64Mi
ingester-whizard	StatefulSet	2	receive	4	16Gi	50m	64Mi	为软租户单独配置
ingester-whizard-(local/remote)-auto-0	StatefulSet	2	receive	4	16Gi	50m	64Mi	将根据租户数量自动扩展
compactor-whizard-remote-xxx	StatefulSet	1	receive	4	16Gi	50m	64Mi	将根据租户数量自动扩展
store-whizard-remote-xxx	StatefulSet	2	receive	4	16Gi	50m	64Mi	将根据租户数量自动扩展
query-frontend-whizard	Deployment	2	query-frontend	2	4Gi	50m	64Mi
query-whizard	Deployment	3	query	2	4Gi	50m	64Mi
			proxy	2	4Gi	50m	64Mi	sidecar 容器
router-whizard	Deployment	2	receive	2	4Gi	50m	64Mi
ruler-<tenant>-0	StatefulSet	1	rule	2	4Gi	50m	64Mi	与 Whizard Tenant 一一对应
			write-proxy	2	4Gi	50m	64Mi	sidecar 容器
			config-reloader	100m	128Mi	100m	128Mi	sidecar 容器
whizard-adapter	Deployment	1	adapter	200m	256Mi	100m	128Mi
whizard-agent-proxy	Deployment	1	whizard-agent-proxy	2	4Gi	50m	64Mi
whizard-controller-manager	Deployment	1	manager	200m	512Mi	100m	128Mi

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%