Bug 1168709 - Hosted Engine VM is listed as paused after upgrading from 3.4.4 to 3.5.1 snapshot
Summary: Hosted Engine VM is listed as paused after upgrading from 3.4.4 to 3.5.1 snap...
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: oVirt
Classification: Retired
Component: vdsm
Version: 3.5
Hardware: Unspecified
OS: Unspecified
unspecified
high
Target Milestone: ---
: 3.5.1
Assignee: Francesco Romani
QA Contact: Gil Klein
URL:
Whiteboard: virt
Depends On:
Blocks: oVirt_3.5.1_tracker
TreeView+ depends on / blocked
 
Reported: 2014-11-27 15:59 UTC by Sandro Bonazzola
Modified: 2016-02-10 19:47 UTC (History)
11 users (show)

Fixed In Version: ovirt-3.5.1_rc1
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2015-01-21 16:03:02 UTC
oVirt Team: Virt
Embargoed:


Attachments (Terms of Use)
screenshot (119.79 KB, image/png)
2014-11-27 15:59 UTC, Sandro Bonazzola
no flags Details
log collector report (13.08 MB, application/x-xz)
2014-11-27 16:12 UTC, Sandro Bonazzola
no flags Details


Links
System ID Private Priority Status Summary Last Updated
oVirt gerrit 36021 0 master MERGED vm: ensure graphics configuration data in recovery Never
oVirt gerrit 36065 0 ovirt-3.5 MERGED vm: ensure graphics configuration data in recovery Never

Description Sandro Bonazzola 2014-11-27 15:59:38 UTC
Created attachment 962147 [details]
screenshot

On a working 3.4.4 CentOS system I upgraded the host to 3.5.1 snapshot and hit bug #1168689 and bug #1168695

Following http://www.ovirt.org/Hosted_Engine_Howto#Upgrade_Hosted_Engine

I did
 vdsm-tool configure --module libvirt --force
 service vdsmd start
 service ovirt-ha-broker restart && service ovirt-ha-agent restart
 and updated to 3.5 the cluster and the datacenter.

The engine is up and running, the host looks ok, the hosted engine vm is listed as paused.

Comment 1 Sandro Bonazzola 2014-11-27 16:12:58 UTC
Created attachment 962160 [details]
log collector report

Comment 2 Sandro Bonazzola 2014-11-28 10:48:58 UTC
Reproducible on F20 hosts and F19 VM too.

Comment 3 Sandro Bonazzola 2014-12-09 10:40:34 UTC
Michal can you take a look?

Comment 4 Sandro Bonazzola 2014-12-09 10:42:16 UTC
Oved, can you take a look too?

Comment 5 Oved Ourfali 2014-12-09 11:40:00 UTC
(In reply to Sandro Bonazzola from comment #4)
> Oved, can you take a look too?

I'll let Michal have an additional look, as it seems like the host was non-responsive, and someone confirmed it was rebooted, and once it was up again the VM status was moved from Unknown to Paused... apperently the status was "restored", which makes me wonder why it was Paused in the first place.

If we have communication to the host, why can't we get the real status of the VM?

Comment 6 Oved Ourfali 2014-12-09 11:40:07 UTC
(In reply to Sandro Bonazzola from comment #4)
> Oved, can you take a look too?

I'll let Michal have an additional look, as it seems like the host was non-responsive, and someone confirmed it was rebooted, and once it was up again the VM status was moved from Unknown to Paused... apperently the status was "restored", which makes me wonder why it was Paused in the first place.

If we have communication to the host, why can't we get the real status of the VM?

Comment 7 Michal Skrivanek 2014-12-09 12:31:15 UTC
seems there's an error on VM recovery.

Comment 8 Francesco Romani 2014-12-09 13:32:06 UTC
ok, I think I got it.

Graphics Devices may be not sent by Engine. To cover that VDSM reconstructs them from other VM configuration data - which is safe to do since the information is there, just organized in a different way.

The recovery path skips that step, so the configuration is inconsistent.
Later in the creation path the configuration is assumed to be present (either fixed by VDSM or given by Engine), but it is not, and this makes the domain initialization fail in the last stages.

Patch coming soon.

Comment 9 Sandro Bonazzola 2015-01-15 14:25:42 UTC
This is an automated message: 
This bug should be fixed in oVirt 3.5.1 RC1, moving to QA

Comment 10 Sandro Bonazzola 2015-01-21 16:03:02 UTC
oVirt 3.5.1 has been released. If problems still persist, please make note of it in this bug report.


Note You need to log in before you can comment on or make changes to this bug.