I have no idea if this is a test failure or the test system failure.
+ rm -rf tests/package + mkdir -p tests/package + cp -rp /workDir/workspace/fedora-rawhide-pr-pipeline/logs/default_provisioners.log /workDir/workspace/fedora-rawhide-pr-pipeline/logs/test_subject.qcow2.guest.log /workDir/workspace/fedora-rawhide-pr-pipeline/logs/test_subject.qcow2.qemu.log /workDir/workspace/fedora-rawhide-pr-pipeline/logs/tests.yml-run.txt tests/package/ + cat /workDir/workspace/fedora-rawhide-pr-pipeline/logs/test.log cat: /workDir/workspace/fedora-rawhide-pr-pipeline/logs/test.log: No such file or directory script returned exit code 1
https://src.fedoraproject.org/rpms/python-setuptools/pull-request/16 https://src.fedoraproject.org/rpms/python-virtualenv/pull-request/8 https://src.fedoraproject.org/rpms/python-wheel/pull-request/5 https://src.fedoraproject.org/rpms/python-wheel/pull-request/4 https://src.fedoraproject.org/rpms/python2/pull-request/34
Yet https://src.fedoraproject.org/rpms/python38/pull-request/4 passed :(
cc @cstratak
https://src.fedoraproject.org/rpms/python38/pull-request/4 passed because it does not require /usr/bin/virtualenv
/usr/bin/virtualenv
From @pviktori:
The log says:
TASK [str-common-pkgs : Install test-specific package requirements] ************ [...] failed: [/workDir/workspace/fedora-rawhide-pr-pipeline/images/test_subject.qcow2] (item=/usr/bin/virtualenv) => {"attempts": 5, "changed": false, "item": "/usr/bin/virtualenv", "msg": "No group usr/bin/virtualenv available.", "results": []}
Looks like something thinks /usr/bin/virtualenv in required_packages is a “group”, not a file to be installed?
Replacing /usr/bin/virtualenv with virtualenv workarounds this issue.
virtualenv
Metadata Update from @bookwar: - Issue tagged with: Standard Test Interface
We are still fighting this bug every now and then.
Now with https://jenkins-continuous-infra.apps.ci.centos.org/blue/organizations/jenkins/fedora-rawhide-pr-pipeline/detail/fedora-rawhide-pr-pipeline/1262/pipeline/
It seems that install-by-binary-path is not supported by ansible:
- hosts: localhost tasks: - name: Install packages package: name: - /usr/bin/virtualenv
Results into the above-mentioned error:
fatal: [localhost]: FAILED! => { "changed": false, "msg": "No group usr/bin/virtualenv available.", "results": []}
Is there any disadvantage of using package name instead? Otherwise I'd suggest to close this issue.
This is a regression. It might be a regression in ansible.
Is there any disadvantage of using package name instead?
The package name changed in between Fedora releases.
Ok, I've filed a new ansible bug for this issue: https://bugzilla.redhat.com/show_bug.cgi?id=1710786
@psss I see the above bugzilla got EOL'd and the upstream issue with ansible was closed:
https://github.com/ansible/ansible/issues/66424
Did we ever come to a satisfactory position on this? The last thing I see was you asking the ansible team a question, and then it was just...closed.
I believe this is be fixed once the pipeline start using the newly build STR (https://bodhi.fedoraproject.org/updates/FEDORA-2020-a906f6f60c) in the new build we will install the packages using dnf command instead of ansible module.
I no longer care about this regression as the Fedoras where the package name was different are long EOL now.
Metadata Update from @churchyard: - Issue status updated to: Closed (was: Open)
Login to comment on this ticket.