I've seen a similar issue on OSX with Docker stable v1.13.0. The issue is intermittent and occurs on multiple OSX laptops in the office. When the issue manifests this is the error message:
error pulling image configuration:
Get https://prod-ap-southeast-2-starport-layer-bucket.s3-REGION.amazonaws.com/XXX:
dial tcp: lookup prod-REGION-starport-layer-bucket.s3-ap-southeast-2.amazonaws.com on 192.168.xxx.xxx:53:
no such host
We have been able to reliably fix it by running the following command:
docker run --rm -v /etc/resolv.conf:/rtemp debian:8 bash -c 'echo "nameserver 8.8.8.8" > /rtemp'
Note this command needs to be run every time the docker client is started. The command is changing the docker engine DNS settings to stop proxying through docker client (and thus using your host DNS settings) and instead to resolve DNS queries through a Google DNS server at 8.8.8.8.
I believe the root cause is that the payload size of the DNS responses for the amazon URL changes, and when it is greater than 512 bytes the Docker client DNS proxy cannot handle the response.
I believe the root cause is discussed in this open docker issue but can't be sure : https://github.com/docker/docker/issues/18842