I have found that since upgrading mock on my sole builder for doing
livemedia tasks that these tasks no longer work. Previously, I had
mock-1.4.6-1.fc26 with no problems, but mock-1.4.8-1 causes issues.
Digging around the logs for a failed livemedia task is always fun
because lots of errors are normal, but not fatal. However, I see in
anaconda-anaconda.log messages to the effect that "The group 'core' is
required for this installation. This group does not exist. This is a
fatal error and installation will be aborted." I think the real cause
though is revealed in anaconda-dnf.librepo.log where I see "Curl error
(6): Couldn't resolve host name for http://hostname..." which stems
from the "url --url=http://hostname..." directive in my kickstart.
I see in the release notes for 1.4.7 and 1.4.8 mentions about changes
related to how network connectivity is affected in the build roots, but
those seem relevant only to systemd-nspawn. Since koji-1.15 hasn't
made it's way to CentOS or Fedora repos yet, I'm still on 1.14 and so
necessarily have a tag option "mock.new_chroot : False" set. To
overcome the issues with upgrading to mock-1.4.8, I've also tried
adding "mock.rpmbuild_networking : True" and "mock.use_host_resolv :
True" but still the build root is preventing name resolution.
Any ideas?
Hi all,
I've finally uploaded koji 1.15 to PyPi [1], so give it a shot!
For whom it is good:
Everybody using virtualenv's installations and koji xmlrpc client in
their scripts.
For whom it is not good:
Anyone who wants to install koji servers (hub, web, builders, ...) -
these are not included and we're not planning to release them this way.
Note, that official release channel is still pagure's releases page [2]
and rpms which can be created from these tarballs by running 'make rpm'.
Rpm installation is preferred one.
[1] https://pypi.python.org/pypi/koji/1.15.0
[2] https://pagure.io/koji/releases
[3] https://pagure.io/koji/issue/458
--
Tomas Kopecek <tkopecek(a)redhat.com>
Release Engineering Development, RedHat
Hi,
is there anyone still using old createrepo and mergerepo, or does it
make sense to deprecate it and e.g. in 1.18 remove support for
createrepo and bundled mergerepo and make current option
use_createrepo_c default behaviour?
I haven't seen any problem with current versions of *_c which are 1)
faster 2) we don't need to bundle it and fix bugs there.
https://pagure.io/koji/issue/716
--
Tomas Kopecek <tkopecek(a)redhat.com>
Release Engineering Development, RedHat