What actually happened, I created fresh CentOS 7.2 instance on the same server, so /var/cache/tripleo-quickstart/images became empty. Then piece of code 73 - 75 lines of /home/boris/tripleo-quickstart/roles/libvirt/setup/undercloud/tasks/fetch_image.yml =============================================================== - 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 <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/stabl... Thanks. Boris. <http://artifacts.ci.centos.org/artifacts/rdo/images/mitaka/delorean/stabl...