update jobs to include hystrix from EPEL

Description

oVirt 4.1 requires hystrix yet this package is filtered in jobs which is causing tests to fail:
http://jenkins.ovirt.org/job/ovirt-engine_master_upgrade-from-4.0_el7_merged/378/

http://jenkins.ovirt.org/job/ovirt-engine_master_upgrade-from-4.0_el7_merged/378/artifact/logs/_home_jenkins_workspace_ovirt-engine_master_upgrade-from-4.0_el7_merged_tmp_repo_rpmbuild.log.tgz

error: Failed build dependencies:
hystrix-core is needed by ovirt-engine-4.1.0-0.0.master.20160629135203.gitf1f59e0.el7.centos.noarch
hystrix-metrics-event-stream is needed by ovirt-engine-4.1.0-0.0.master.20160629135203.gitf1f59e0.el7.centos.noarch
Installing /home/jenkins/workspace/ovirt-engine_master_upgrade-from-4.0_el7_merged/tmp_repo/ovirt-engine-4.1.0-0.0.master.20160629135203.el7.centos.src.rpm

Activity

Show:

Eyal Edri December 8, 2016 at 3:06 PM

See the patch is merged, if there are other issues lets open a ticket specific on the issue

Eyal Edri November 28, 2016 at 9:19 AM

Is the issue causing failures still exists?

Former user October 10, 2016 at 8:34 AM

Patch to install packages via puppet as this is affecting newly
installed slaves and causing unpredictable job failures:
https://gerrit.ovirt.org/65291

Barak Korren September 22, 2016 at 7:44 AM

we don`t mean moving the whole job (this cannot be node easilty because there is some crucial stuff needed for engine install missing inside mock, such as systemd), just moving the build part of it to run inside mock_runner.sh and use the same scripts the rest of the world uses to build engine.

Nadav Goldin September 22, 2016 at 7:40 AM

If you mean move the job entirely into standard ci and mock, if I remember correctly there was some limitation on running the upgrade-scripts inside mock, though I'm not familiar enough with the job details. I have an open ticket https://ovirt-jira.atlassian.net/browse/OST-3 to create a new upgrade suite inside ovirt-system-tests using Lago, but I don't ETA on that yet.
- you remember what was the limitation on running the upgrade suites inside mock?

BTW - as a side note - as far as I can tell, we have a limited number of jobs that don't run in standard-ci, a path might worth exploring is having a separate labels & hostgroups for those slaves and applying the jenkins_slave packages on them only. That way 95% of our slaves will be 'thin', and another 5% for legacy jobs, which should disappear eventually.

Fixed

Details

Assignee

Reporter

Priority

Created June 30, 2016 at 9:28 AM
Updated February 27, 2017 at 8:50 AM
Resolved December 8, 2016 at 3:06 PM