Run ecs task definition has error RESOURCE:GPU AGENT

Question

I run ecs task with gpu container instance

In task definition i add gpu to resourceRequirements with ec2 gpu 
for example when i run task with p2.xlarge value of gpu will be 1

I guess my instance does not have enough remaining gpu. 
Is any way for calculate remaining gpu for create task definition?

Thanks!!

Osvaldo Marte · Accepted Answer

You can use the AWS CLI to describe your ECS container instances and extract the relevant information. For example you can do this:

1. Use the AWS CLI to list the ARNs of the container instances in your cluster.
   ```sh
   aws ecs list-container-instances --cluster 
   ```

2. Use the ARNs from the previous step to get detailed information about each instance.
   ```sh
   aws ecs describe-container-instances --cluster  --container-instances   ...
   ```

3. From the output, extract the total and remaining GPUs for each instance. Look for the `registeredResources` and `remainingResources` sections, specifically for entries where the `name` is `GPU`.

4. Subtract the number of GPUs in use from the total number of GPUs to determine the remaining GPUs available for allocation.

> Here you can see an example of the [output](https://docs.aws.amazon.com/cli/latest/reference/ecs/describe-container-instances.html#examples).

Run ecs task definition has error RESOURCE:GPU AGENT

Relevant content