ECS Cluster CPU Reservation Saturation

critical

Resource ContentionUpdated Nov 6, 2025

High CPU reservation in ECS clusters leads to task placement failures where new tasks remain pending indefinitely, preventing service scaling and causing cascading latency issues.

Sources

AWS ECS metrics | Vector documentationvector.dev

Ecs Fargate Memory And CPU - TechBloatwww.techbloat.com

Catch and remediate ECS issues faster with default monitors and the ECS Explorer | Datadogwww.datadoghq.com

Key Amazon ECS Metrics to Monitor - Site24x7 Learnsite24x7.com

Technologies:

Amazon ECSThe root cause of this issue originates in Amazon ECS

DockerSymptoms of this issue are visible in Docker metrics and logs

How to detect:

Monitor CPUReservation metric approaching 100% of cluster capacity. Symptoms include tasks stuck in PENDING state, service events showing placement failures, and inability to scale services despite low actual CPU utilization.

Recommended action:

Compare CPUReservation vs CPUUtilization across services to identify over-reserved containers. Adjust container-level CPU limits in task definitions to match actual usage. Scale cluster capacity if utilization justifies reservation levels. Use ECS Explorer to identify which services are over-reserving relative to usage.