brentley/tensorflow-container

Example code for deploying GPU workloads on ECS

tensorflow-container's Issues

Parameter validation failed. "resourceRequirements", must be one of ...

I try to reproduce your code but cannot register a task with resourceRequirements

            "resourceRequirements": [
                {
                    "type": "GPU",
                    "value": "1"
                }
            ],

P.S without resourceRequirements lines task can be registered easely. and I also use us-east-2 region

When I try to register the code with your task definition json file I get the error:
$ aws ecs register-task-definition --cli-input-json file://gpu-1-taskdef.json

Parameter validation failed:
Unknown parameter in containerDefinitions[0]: "resourceRequirements", must be one of: name, image, repositoryCredentials, cpu, memory, memoryReservation, links, portMappings, essential, entryPoint, command, environment, mountPoints, volumesFrom, linuxParameters, secrets, hostname, user, workingDirectory, disableNetworking, privileged, readonlyRootFilesystem, dnsServers, dnsSearchDomains, extraHosts, dockerSecurityOptions, interactive, pseudoTerminal, dockerLabels, ulimits, logConfiguration, healthCheck, systemControls

I also tried to move resourceRequirements to other subfields but it didn't help.

From docs I can see that it's valid parameter.
https://docs.aws.amazon.com/cli/latest/reference/ecs/register-task-definition.html

Maybe something was changed with resourceRequirements? Can you recheck that all works great.