Understanding memory in ECS EC2 task

Question

How can I tell an ECS container that it cannot use more than 10GB of memory?

I don't give the task any limits because I need alltogether in all container more than 10 CPU units and EC2 has a limit of 10 CPUs in the task definition.

In the container definition I give the container a memoryReservationMiB of 10GB and cpu of 6. But the container doesn't seem to care. I want to prevent the container from terminating if too much memory is used, so I don't set memoryLimitMiB but only memoryReservationMiB.

Unfortunately the application crashes in the ECS EC2 task with a memory error, but in EKS the container runs through with the same settings.

```
          resources:
            requests:
              cpu: "4"
              memory: "4Gi"
            limits:
              cpu: "6"
              memory: "10Gi
```

Is the approach wrong?

Jordan Sandri · Answer

Hello,

When creating a Task Definition on ECS, you can set the **CPU** and **memory allocation** in the **Resource allocation limits** field (see screenshot below). If you've set CPU to 6 vCPU and **Memory soft limit** to 10GB it should work. If your container is crashing, have you checked if your EC2 instance can actually handle more than 10GB of memory? Also, do you know how much memory your task might actually need at peak usage?

Hope it helps :)

![Enter image description here](/media/postImages/original/IMYxMogXk-RTCLXzJqXoohbA)

Understanding memory in ECS EC2 task

Add your answer

Relevant content