[ https://ovirt-jira.atlassian.net/browse/OVIRT-1840?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=35638#comment-35638 ]
Looking at timestamps of when docker was installed, seems that the slave was broken during this build: http://jenkins.ovirt.org/job/vdsm_master_check-merged-el7-x86_64/3039/console
07:49:56 Running transaction 07:49:57 Updating : container-storage-setup-0.8.0-3.git1d27ecf.el7.noarch 1/12 07:49:57 Updating : 2:oci-umount-2.3.0-1.git51e7c50.el7.x86_64 2/12 07:50:35 Updating : 2:container-selinux-2.33-1.git86f33cd.el7.noarch 3/12 07:50:36 Updating : 2:docker-common-1.12.6-68.gitec8512b.el7.centos.x86_64 4/12 07:50:38 Updating : 2:docker-client-1.12.6-68.gitec8512b.el7.centos.x86_64 5/12 07:50:39 Updating : 2:docker-1.12.6-68.gitec8512b.el7.centos.x86_64 6/12 13:49:32 Cleanup : 2:docker-1.12.6-48.git0fdc778.el7.centos.x86_64 7/12 Build timed out (after 360 minutes). Marking the build as failed.
This explains the leftover yum processes on the system, which can block further yum installs or even theoretically corrupt RPMDB if docker decides to unfreeze for some reason.
jobs freeze due to unresponsive docker
Key: OVIRT-1840 URL: https://ovirt-jira.atlassian.net/browse/OVIRT-1840 Project: oVirt - virtualization made easy Issue Type: Task Reporter: Evgheni Dereveanchin Assignee: infraQuite often do I see jobs stuck at various stages for hours that seem related to docker. Example: http://jenkins.ovirt.org/job/ovirt-engine_master_build-artifacts-fc26-x86_64/610/console There's multiple docker commands stuck on the slave (will post in the next comment) so it seems to be deadlocked. Opening ticket to investigate which step exactly is causing this and possible ways of resolving. The job in question doesn't even use docker so shouldn't suffer if this happens.
— This message was sent by Atlassian Jira (v1001.0.0-SNAPSHOT#100075)