Hello. I am attempting to deploy to AWS ECS in my company’s account. I have configured the

docker-compose.yml

file and when I run

docker compose up

the dagit, daemon, and grpc service create locally. However when I attempt to run the

up

command with the ecs context to deploy to aws, my grcp service fails to create. The moment this happens, all resource automatically delete, and it seems I I lose all information on the failure. Where do I look for failure logs in light of this behavior? Is there a way to stop the automatic deletion on failure since I can run

down

?

Hey Aaron - our resident expert on the docker-compose ECS integration is out this week but I will do my best in his absence. Under the hood I believe this integration creates a Cloudformation template and creates that as per https://docs.docker.com/cloud/ecs-integration/#run-a-compose-application Is there any more information about the failure in the Cloudformation console when this happens?

No. And I’m unable to investigate the failure because all of the resources are automatically deleted, including the Cloudformation and Log group

What about converting it to a Cloudformation template using "docker compose convert" and then running it just within Cloudformation?

I’ll try that! Didn’t know it was an option. Not a DevOps person.

Yeah we've been thinking about making the official dagster ECS example just use Cloudformation directly instead - the docker-compose integration is nice for getting something up quickly but seems to have a lot of rough edges like this

@daniel That helped a bunch. I was able to diagnose and solve the issue! Thank you!