- name: Get image
      command: >
        curl -sf -C- -o _{{ image.name }}.{{ image.type}} {{ image.url }}
      args:
        chdir: "{{ image_cache_dir }}"
      register: curl_result
      until: curl_result.rc not in [18, 56]
      retries: 20
      delay: 5
================================================================
doesn't loop for me any longer ( as happened before, since late May 2016 )

It sits on first curl invocation with spead 2.85MB/min  , e.g. 10-12 hr for download 2.5 GB.
Due to "-C-" option used by curl  in about several days via restarting `bash quickstart $VIRTHOST`
on regular basis undercloud.qcow2 will be downloaded, it didn't happen just 2-3 days ago
probably due to using cached copy of undercloud.qcow2.
However , that was a point of time in the past ( around 05/25 ) when I picked  up
undecloud.qcow2 the first time with no problems.
I also believe that all what I did  was just the help to solve the issue with telemetry services failure :-
https://www.redhat.com/archives/rdo-list/2016-June/msg00036.html

Appears that my ISP has problem with only one site's downloading speed .
This site is http://artifacts.ci.centos.org
Link to to file   http://artifacts.ci.centos.org/artifacts/rdo/images/mitaka/delorean/stable/undercloud.qcow2

Thanks.
Boris.