Jenkins failure - perhaps bad permissions

Description

http://jenkins.ovirt.org/view/oVirt%20system%20tests/job/ovirt-system-tests_manual/1985/

10:53:01 ERROR: Error fetching remote repo 'origin'10:53:01
hudson.plugins.git.GitException: Failed to fetch from
git://gerrit.ovirt.org/ovirt-system-tests.git*10:53:01* at
hudson.plugins.git.GitSCM.fetchFrom(GitSCM.java:825)10:53:01 at
hudson.plugins.git.GitSCM.retrieveChanges(GitSCM.java:1092)10:53:01
at hudson.plugins.git.GitSCM.checkout(GitSCM.java:1123)10:53:01 at
org.jenkinsci.plugins.multiplescms.MultiSCM.checkout(MultiSCM.java:143)10:53:01
at hudson.scm.SCM.checkout(SCM.java:495)10:53:01 at
hudson.model.AbstractProject.checkout(AbstractProject.java:1202)10:53:01
at hudson.model.AbstractBuild$AbstractBuildExecution.defaultCheckout(AbstractBuild.java:574)10:53:01
at jenkins.scm.SCMCheckoutStrategy.checkout(SCMCheckoutStrategy.java:86)10:53:01
at hudson.model.AbstractBuild$AbstractBuildExecution.run(AbstractBuild.java:499)10:53:01
at hudson.model.Run.execute(Run.java:1724)10:53:01 at
hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:43)10:53:01 at
hudson.model.ResourceController.execute(ResourceController.java:97)10:53:01
at hudson.model.Executor.run(Executor.java:421)10:53:01 Caused by:
hudson.plugins.git.GitException: Command "git clean -fdx" returned
status code 1:10:53:01 stdout: Removing
basic-suite-master/LagoInitFile*10:53:01* Removing
basic-suite-master/extra_sources*10:53:01* Removing
basic-suite-master/reposync-config.repo_repoclosure*10:53:01* Removing
basic-suite-master/test-scenarios/000_check_repo_closure.pyc*10:53:01*
Removing basic-suite-master/test-scenarios/001_initialize_engine.pyc*10:53:01*
Removing basic-suite-master/test-scenarios/002_bootstrap.pyc*10:53:01*
Removing basic-suite-master/test-scenarios/003_00_metrics_bootstrap.pyc*10:53:01*
Removing basic-suite-master/test_utils/_init_.pyc*10:53:01* Removing
basic-suite-master/test_utils/constants.pyc*10:53:01* Removing
basic-suite-master/test_utils/network_utils_v4.pyc*10:53:01* Removing
exported-artifacts/10:53:01 Removing mock_logs.LjHRcpGf/10:53:01
Removing mock_script_sid.x4jv1R*10:53:01* Removing
mocker-epel-7-x86_64.el7.cfg*10:53:01* 10:53:01 stderr: warning:
failed to remove .pki/10:53:01 warning: failed to remove
deployment-basic-suite-master/default/uuid*10:53:01* warning: failed
to remove deployment-basic-suite-master/default/id_rsa
...


Didi

Activity

Show:

Barak Korren June 5, 2018 at 10:29 AM

I know that signal handling in Jenkins is a difficult task. Maybe let's change the job to wipe the workspace before running git?

That would destroy all the caches we worked very hard to keep around, including the Git cache itself which cab be quite significant.

As I already commented elsewhere, this is not an easy problem to solve in our current architecture:
For V1 Jobs - because Jenkins freestyle jobs do not allow you to easily add functionality before the git cloning is performed.
For V2 jobs - pipelines to not allow you to embed arbitrary shell scripts in the jobs so you need to clone the repo before you can run any meaningful preparation steps.

This issue is essentially an aspect of what we call "the chicken and egg" issue where we have to prepare the Jenkins slave before we can clone any code to it, but we need the code to do the preparation. Our current direction for solving this long-term is to move the slave preparation steps out from the jobs ()

Gal Ben Haim June 5, 2018 at 9:44 AM

I don't see any valid solution from OST side.
any idea how we can solve it?
I know that signal handling in Jenkins is a difficult task. Maybe let's change the job to wipe the workspace before running git?

Eyal Edri June 4, 2018 at 1:59 PM

is there a different card for tracking the proposed long-term solution?

Gal Ben Haim January 9, 2018 at 5:04 PM

The issue was solved, and we are working on a long-term solution.

Shirly Radco January 9, 2018 at 2:12 PM

Any update on this?

Gal Ben Haim January 9, 2018 at 1:19 PM

The lago prefix is created by root inside mock, build http://jenkins.ovirt.org/job/ovirt-system-tests_manual/1984/ was aborted, and the cleanup has not been called (we aware about this issue), and then the next build failed to clean OST repo (with git clean).

is working on a solution for this issue.

Cannot Reproduce

Details

Assignee

Reporter

Priority

Created January 9, 2018 at 11:04 AM
Updated August 29, 2019 at 2:12 PM
Resolved December 24, 2018 at 9:40 AM