As described here ( https://bugzilla.redhat.com/show_bug.cgi?id=1514511 ) there is an issue here systemd would stop the docker service incorrectly. Often seeming to occur when a container was in a restart state. A patch has been merged to address this: https://code.engineering.redhat.com/gerrit/#/c/124023/
@Omri: I believe that as well as a hard reboot, you also need a container that's failing and being restarted, to properly trigger this issue.
Verified with puppet-tripleo-7.4.3-10.el7ost.noarch unable to reproduce the race issue by running overcloud reboot loop while one of the containers is in "Restarting" status . steps: (1) caused horizon container to be stuck in 'restarting status' [root@overcloud-controller-1 ~]# docker ps | grep horizon 6047011009b6 192.168.0.1:8787/rhosp12/openstack-horizon-docker:12.0-20171128.1 "kolla_start" 29 minutes ago Restarting (2) About a min horizon (changed the docker start command pointing to different ports ) (2) ran "Killer Reboot test case" - 5 X reboot , wait 5 minutes sleep after each reboot. results : docker is still up and running on the OC nodes. while the horizon container is still stuck in restarting.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2017:3462