Description of problem: copr cannot download SRPMS from an IPv6 only host Version-Release number of selected component (if applicable): How reproducible: always Steps to Reproduce: 1.publish SRPM to host accessible only by IPv6 e.g. http://buildserver.therobinsonfamily.net/SRPMS/ 2.???? 3.Fail Actual results:Build fails with: Error Downloading http://buildserver.therobinsonfamily.net/SRPMS/tty-solitaire-git-4.20150530git343a55ca.fc22.src.rpm: HTTPConnectionPool(host='buildserver.therobinsonfamily.net', port=80): Max retries exceeded with url: /SRPMS/tty-solitaire-git-4.20150530git343a55ca.fc22.src.rpm (Caused by <class 'socket.error'>: [Errno 101] Network is unreachable) Expected results: SRPM to be built Additional info:
Copr is deployed as instance on OpenStack. And OpenStack still does not fully support IPv6. This will take long. Probably before OS will start supporting IPv6, we will implement uploading of SRPM.
*** Bug 1812077 has been marked as a duplicate of this bug. ***
Copr now lives mostly in AWS, and we should be able to configure ipv6 there. This needs to be done in cooperation with fedora infra people because we (copr team) don't have permissions to configure subnets. This should resolve also the `repo` files downloaded by `dnf copr enable`.
*** Bug 1900300 has been marked as a duplicate of this bug. ***
Not only frontend, but backend needs fix too, from bug 1900300: > Additional info: > while 'https://download.copr.fedorainfracloud.org' is reachable via ipv6 > trying to access the real contain (the location epel-8-$basearch/ in my case) > redirect to https://copr-be.cloud.fedoraproject.org which not reachable via ipv6 as well. The CDN redirects DNS queries to copr-be.cloud.* if particular file isn't cached.
AWS environment allowed us to setup IPv6 networking. So copr-cli (communicates with copr-frontend should work) same as downloading the built results (downloads from copr-backend) should work.