CloudWatch Dashboard


1. Dashboard Rows#

Row CloudWatch Metrics Statistic Display
Overview RunningTaskCount, DesiredTaskCount Average stat / time series by service
Overview CPUUtilization Average time series by service
Overview MemoryUtilization Average time series by service
Overview, if HTTP behind ALB ALB RequestCount Sum request rate
Overview, if HTTP behind ALB ALB HTTPCode_Target_5XX_Count, HTTPCode_ELB_5XX_Count Sum target error and ALB-layer error
Overview, if HTTP behind ALB ALB TargetResponseTime p95 / p99 request latency
Drilldown PendingTaskCount Average time series by service
Drilldown PendingTaskCount from ECS/ContainerInsights Average deployment blocked / capacity shortage signal
Capacity DesiredTaskCount, RunningTaskCount Average trend
Debug service-level pending / running task count Average deployment troubleshooting
Debug PendingTaskCount with service events and logs Average deployment / scheduling / capacity issue investigation

2. Notes#

ECS CloudWatch metrics 只能说明 service / task 资源状态。

ECS namespace:
    CPUUtilization
    MemoryUtilization
    RunningTaskCount
    DesiredTaskCount
    PendingTaskCount
    PendingTaskCount, from ECS/ContainerInsights

ALB metrics, if service is HTTP behind ALB:
    RequestCount
    HTTPCode_Target_5XX_Count
    HTTPCode_ELB_5XX_Count
    TargetResponseTime

App metrics, recommended:
    http_requests_total
    http_request_duration_seconds
    error counter by route / status / exception

结论:
    可以从 CloudWatch 看 request / error / latency，但通常不是 ECS namespace。
    如果服务 behind ALB，就看 AWS/ApplicationELB。
    如果要看真正应用级 request / error / latency，最好应用自己暴露 Prometheus metrics。
Backward YACE Dashboard