[one-users] VM Sum isn't consistent

[Duverne, Cyrille]

Hello Ruben,

Thanks for this feedback.

No problem with the inconvenience, that's a bit tricky but not making
the infra failing :)

If I understood well the upgrade process, it's less impactant than 3.2
to 3.6 and can be dealt via packages.

Does this mean that my actual config files will be automatically
upgraded ? Or do I have to update the new ones with the old config ?

To be 100% sure, I don't have to stop my running VMs just be sure that
they all are in a final state ?

Thanks in advance.
Cyrille

"Imagination is more important than Knowledge"
Albert Einstein

Mercredi 24/10/2012 à 10:22 Ruben S. Montero a écrit:

Hi

Yes, we have included some modifications to check these corner cases
in OpenNebula 3.8.. We are also having problems to reproduce this
issue, but it seems to happen when deleting a VM and some race
condition with the VM moving to another state at the same time. So we
are not 100% sure this is completely fixed in 3.8.

After some discussions we have decided to include some additional
logic to prevent this to happen (keep some extra relationships in the
objects). As a bonus we'll get the VMs IDs that are running on a host
or the  specific VMs that using an image directly from the 'onehost
show' and 'oneimage show' commands; and sunstone views.

Now, how to fix this. I'd recommend to:

0.- Stop OpenNebula
1.- BACK UP, BACK UP and BACK UP your database
2.- Upgrade to OpenNebula 3.8 (may probably solve this from happen
again)
3.- Use the new fsck option of onedb that will fix the figures for the
images and hosts. [2]
4.- Restart your services.

[1] http://opennebula.org/documentation:rel3.8:upgrade
[2] http://opennebula.org/documentation:rel3.8:onedb#onedb_fsck

We'll start working on this ASAP and provide a patch for the
additional checks. 

Thanks for your feedbacl and really sorry for the inconvenience :(

Best

Ruben 

On Tue, Oct 23, 2012 at 10:15 PM, Giovanni Toraldo  wrote:

Hi,

2012/10/23 Duverne, Cyrille :
>> How could we solve this ?
>> Bug or feature ?

I am facing this bug as well, but I am not able to reproduce it
consistently.

Last time happened while testing a VMWare infrastructure with ON 3.6,
I usually correct the wrong VM count directly on the database after
I've finished tests.

I think there is some corner case that lead to multiple subtractions
in the host counter for the same VM (after a FAIL, maybe).

--
Giovanni Toraldo
http://gionn.net
_______________________________________________
Users mailing list
Users at lists.opennebula.org
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org

-- 
Ruben S. Montero, PhD
Project co-Lead and Chief Architect
OpenNebula - The Open Source Solution for Data Center Virtualization
www.OpenNebula.org [1] | rsmontero at opennebula.org | @OpenNebula



Links:
------
[1] http://www.OpenNebula.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20121024/c313da7d/attachment-0002.htm>