[ovirt-users] Affinity/HA for a set of VMs

TranceWorldLogic . tranceworldlogic at gmail.com
Wed Dec 14 10:10:48 UTC 2016


Hi,

After looking into some more document, I think storage level fencing is
supported in ovirt via gluster.
Can some one correct me on this ?

Thanks,
~Rohit

On Wed, Dec 14, 2016 at 2:32 PM, TranceWorldLogic . <
tranceworldlogic at gmail.com> wrote:

> Hi Doron,
>
> Yes, I am referring to storage level fencing like drbd/pacemaker tool.
>
> Thanks,
> ~Rohit
>
>
> On Wed, Dec 14, 2016 at 1:42 PM, Doron Fediuck <dfediuck at redhat.com>
> wrote:
>
>>
>>
>> On Wed, Dec 14, 2016 at 9:23 AM, Yedidyah Bar David <didi at redhat.com>
>> wrote:
>>
>>> On Wed, Dec 14, 2016 at 9:02 AM, TranceWorldLogic .
>>> <tranceworldlogic at gmail.com> wrote:
>>> > Hi,
>>> >
>>> > I was trying to explore more about fencing option supported in ovirt.
>>> > But getting lost in documents.
>>> >
>>> > My requirement is to fence at VM level rather than host level.
>>>
>>> I am pretty certain that this is not called fencing. Fencing is for
>>> hosts.
>>> And it's meant for _stopping_ them, not starting.
>>>
>>> > e.g let assume VM1.1, VM1.2,VM1.3 are running on host1 and VM2.1,VM2.2
>>> VM2.3
>>> > running on host2. Suppose due to some error only VM1.1 goes down [Note:
>>> > VM1.2 and VM1.3 in running state] then VM2.1 must come up.
>>> >
>>> > Can I get to know whether such functionality is supported by ovirt ?
>>> > If yes, would you please explain also how it work
>>> >           or would you share refernece for me to refer and understand
>>> it ?
>>> > if yes, is it configurable by python sdk ?
>>>
>>> No idea, but it might be possible using affinity rules and/or HA.
>>> Changing the subject, to try to attract more relevant responses,
>>> and adding Tomer.
>>>
>>> Best,
>>> --
>>> Didi
>>>
>>>
>>>
>> Hi,
>> are you referring to storage based fencing?
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20161214/fb2601f7/attachment-0001.html>


More information about the Users mailing list