i have a question, as I’m in a similar horror movie :-)

Wait, I mean :-(

EBS volumes are AZ-limited, and I see you have one per AZ. When an ECS task definition changes, and ASG is doing its magic, performing a rolling update, I don’t see how there will be any unattached volumes to find. ASG will want to bring a new instance up and have it signal readiness (via cfn-signal) before the old one is terminated. Meanwhile, the old one will be attached to the EBS volume that the new one will be hunting for. Won’t there be a deadlock?