[Users] hosted-engine rebooting in the middle of setup (was: [vdsm] [ANN] oVirt 3.4.0 Release Candidate is now available)

From: "Darrell Budic" <darrell.budic(a)zenfire.com&gt; To: "Sandro Bonazzola" <sbonazzo(a)redhat.com&gt; Cc: announce(a)ovirt.org, "engine-devel" <engine-devel(a)ovirt.org&gt;, "arch" <arch(a)ovirt.org&gt;, Users(a)ovirt.org, "VDSM Project Development" <vdsm-devel(a)lists.fedorahosted.org&gt; Sent: Saturday, March 1, 2014 1:56:23 AM Subject: Re: [vdsm] [Users] [ANN] oVirt 3.4.0 Release Candidate is now available Started testing this on two self-hosted clusters, with mixed results. There were updates from 3.4.0 beta 3. On both, got informed the system was going to reboot in 2 minutes while it was still installing yum updates. On the faster system, the whole update process finished before the 2 minutes were up, the VM restarted, and all appears normal. On the other, slower cluster, the 2 minutes hit while the yum updates were still being installed, and the system rebooted. It continued rebooting every 3 minutes or so, and the engine console web pages are not available because the engine doesn’t start. it did this at least 3 times before I went ahead and reran engine-setup, which completed successfully. The system stopped restarting and the web interface was available again. A quick perusal of system logs and engine-setup logs didn’t reveal what requested the reboot. That was rather impolite of something to do that without warning :) At least it was recoverable. Seems like scheduling the reboot while the yum updates were still running seems like a poor idea as well.