Can't do EL8 builds due to modular packages
by Mátyás Selmeci
Hi,
I've been trying to do EL8 builds on our Koji instance
(koji.opensciencegrid.org) but installing software into the
buildroot fails with error messages like
No available modular metadata for modular package 'subversion-1.10.2-1.module_el8.0.0+45+75bba4f4.x86_64', it cannot be installed on the system
The build machine is CentOS 8 with koji-builder
1.17.0 and mock 1.4.21. The hub is CentOS 7 with koji 1.17.0.
This is what the build tag looks like:
Tag: dist-el8-build [663]
Arches: x86_64
Groups: build, srpm-build
Required permission: 'admin'
Tag options:
Targets that build into this tag:
dist-el8-build (dist-el8-build, repo#71484: 2020-04-22 17:26:22.270397)
This tag is a buildroot for one or more targets
Current repo: repo#71484: 2020-04-22 17:26:22.270397
Targets that build from this tag:
dist-el8-build
External repos:
9 dist-el8-centos-PowerTools (http://mirror.centos.org/centos-8/8/PowerTools/$arch/os/)
10 dist-el8-centos-BaseOS (http://mirror.centos.org/centos-8/8/BaseOS/$arch/os/)
11 dist-el8-centos-AppStream (http://mirror.centos.org/centos-8/8/AppStream/$arch/os/)
(I'm using the main upstream repo just to rule out any mirror issues
on our end.)
I also attached mock_output.log.
How can I fix this?
Thanks,
-Mat
--
Mátyás (Mat) Selmeci
Open Science Grid Software Team / Center for High-Throughput Computing
University of Wisconsin-Madison Department of Computer Sciences
3 years
Regression in 1.21 (admin can't force tag build)
by Tomas Kopecek
We've found a bug which was introduced in 1.21 that admin now can't
override policy via --force when tagging a build. There is an issue [1] and
fix proposal [2] for that. Anyway, if you're depending on such behaviour do
have it in mind when upgrading your hub to 1.21 (and apply this patch
also). It shouldn't affect anything else (other --force usage) as this is
the only task running from builder and not directly via API.
This is an early warning - we're now testing that fix, so have an eye on
the "testing-done" label before we can be sure that it works and it doesn't
break anything else.
[1] https://pagure.io/koji/issue/2202
[2] https://pagure.io/koji/pull-request/2203
--
Tomas Kopecek <tkopecek(a)redhat.com>
RHEL Build Development, RedHat
3 years
koji bugs in bugzilla.redhat.com
by Kevin Fenzi
Greetings koji-devel.
I've been (mostly) handling updates for koji in fedora/epel for a while
now. I noticed a while back that there's a number of bugs filed against
koji in bugzilla.redhat.com for fedora/epel that are... really upstream
bugs. RFE's or asking to change something that I am definitely not going
to change downstream.
Does anyone object to me going through them and asking reporters for all
the non fedora/epel packaging issues to refile upstream on
https://pagure.io/koji ?
Here's a list:
1806193 NEW - Mike McLean - koji-1.17 is not compatibile with python 2.6.6
1734883 NEW - Mike McLean - koji builds fail on s390x sometimes, SRPM cannot be unpacked: cpio: read failed - Inappropriate ioctl for device
1641191 NEW - Mike McLean - Please make pm_request available (and working) in koji
1537197 NEW - Mike McLean - [RFE] supply egg-info in python2-koji and python3-koji packages
1497923 NEW - Dennis Gilmore - koji-web requires mod_auth_gssapi but that is not available in RHEL6 or EPEL6
1481983 NEW - Mike McLean - koji: Insecure host name canonicalization breaks Kerberos security
1478912 NEW - Mike McLean - koji resubmit: TypeError: can't multiply sequence by non-int of type 'str'
1276187 NEW - Mike McLean - RFE: manage ostree options in kickstarts when doing image builds.
1130233 NEW - Mike McLean - koji: Immediate scratch build purging
1096480 NEW - Dennis Gilmore - Koji uses "--target arm" and thus causes wrong build failures
1002227 NEW - Dennis Gilmore - kojid: On stage createLiveCD, livecd-creator freezes when working in mock chroot
915440 NEW - Mike McLean - RFE: Always keep latest build log
748824 NEW - Mike McLean - RFE: mock mount plugin support
692030 NEW - Dennis Gilmore - traceback within the web pages if the repo is already deleted
688578 NEW - Mike McLean - RFE: koji web interface should not show email addresses from changelog in clear text
571484 NEW - Mike McLean - RFE: provide direct URL to logfile when a failure occurs
561998 NEW - Mike McLean - RFE: koji database connectivity needs to be expressly testable and to suggest the 'next' fix needed
Thanks!
kevin
3 years, 1 month
Koji 1.21.0 release
by Tomas Kopecek
Koji 1.21.0 is out. Thanks to everyone who contributed (17 authors)!
You can read release notes here:
https://docs.pagure.org/koji/release_notes/release_notes_1.21/
Highlights:
* sidetag plugin is now part of koji (with new features already)
* clone-tag can now "add" content to destination tag
* download-build resumes failed downloads
* optional rebuildSRPM skipping for scratch builds
* kojid is now not DoSed by big failed builds
* new policy for dist-repo
* kojira can monitor external repositories
* various speed improvements again
* 73 pull requests ( + 18 in 1.20.1)
You can view the 1.21 roadmap at https://pagure.io/koji/roadmap/1.21
For the current roadmap, see https://pagure.io/koji/roadmap
You can download this and other releases at https://pagure.io/koji/releases
--
Tomas Kopecek <tkopecek(a)redhat.com>
RHEL Build Development, RedHat
3 years, 1 month
Running mock in a buildroot from postSCMCheckout allback
by Adam Saleh
Hi all,
we are working on a koji plugin that would automate some of the up-keep
with changelog and release field in spec-files.
ec-files.
As we are testing this out in stg koji instance, we realized it would make
our life easier, if we just could call i.e. rpmspec on the files in the
buildroot directly, as opposed to having just the plain srcdir that gets
passed to the callback directly.
As far as we see, in the buildSRPMfromSCM task, the buildroot is created,
but doesn't get passed into any of the callbacks.
Any ideas how we could do this?
Adam
3 years, 1 month