ECS/EC2 instances don't download ECR image from task definition


I uploaded my app image to a private repo on ECR, when I start my EC2 instances (autoscale group for ECS service) they hang and eventually fail with "Task failed to start". I have connected to the instances using ssh, and from what I can tell docker never tries to download the image from ECR. I have confirmed that the task definition has the ECR image url defined in the container definitions -> image field, and that the image listed there exists using docker manifest inspect. Anyone have any idea why this could be happening?

1 Answer

The task could fail owing to a variety of reasons. If you go under ECS --> Cluster --> service --> Events, you should see why that task failed.

profile pictureAWS
answered a year ago
  • There are no events for the failures. The only events shown there are 'service ... has started 1 task'. When I view aws ecs describe-tasks --tasks [task_id] the stopped reason is as follows:
    "stopCode": "TaskFailedToStart", "stoppedAt": "2022-12-20T07:44:46.292000-08:00", "stoppedReason": "Task failed to start", "stoppingAt": "2022-12-20T07:44:46.292000-08:00",

    There is no evidence that the ecs-agent is even downloading the repo in the first place, but the image is defined in the task definition, and docker manifest inspect verifies that the image exists. It seems like the ecs-agent is just not even starting to process the container logic for the task definition.

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions