Hi,
the solution for this would be either
use gluster or another shared storage
software across all 3 hosts
or go with local storage datacenters.
Am 22.04.2014 20:36, schrieb Tamer Lima:
hello,
I am in trouble
I have 3 servers dedicated to test OVIRT:
01- engine + vdsm (8 cpus, 32GB ram , 2TB HD)
02 - vdsm (8 cpus, 32GB ram , 2TB HD)
03 - vdsm (8 cpus, 32GB ram , 2TB HD)
I want to create cloned virtual machines but in my configuration I can only
save virtual machines on server 01; my configuration refers a DATA DOMAIN
on server 01
All my virtual machines are : 2 cpu , 6 GB ram , 500gb HD and were created
like CLONE
My server 01 is the data domain and all new virtual machine is created, via
NFS, on server 01 , who has 2TB maximum capacity ( the same size of
partition /sda3 = 2TB)
how can I save each virtual machine on a desired vdsm server ?
What I want is :
server 01 - engine + vdsm : 03 virtual machines running and hosted
phisicaly on this host
server 02 - vdsm : 04 virtual machines running and hosted phisicaly
on this host
server 03 - vdsm : 04 virtual machines running and hosted phisicaly
on this host
but I have this :
server 01 - engine + vdsm : 03 virtual machines running and hosted
phisicaly on this host
server 02 - vdsm : 01 virtual machines running on this server BUT
hosted phisicaly on server 01
server 03 - vdsm : none, because my DATA DOMAIN IS FULL (2TB )
How to solve this problem ?
is it possible create one DATA DOMAIN for each VDSM host ? I think this
is the solution but I do not know how to point VMs to be saved on specific
data domain.
thanks
On Fri, Apr 18, 2014 at 4:48 AM, Michal Skrivanek <
michal.skrivanek(a)redhat.com> wrote:
>
> On Apr 17, 2014, at 16:43 , Tamer Lima <tamer.americo(a)gmail.com> wrote:
>
>> hi, thanks for reply
>>
>> I am investigating what is and how "thin virtualization" works
>>
>> Do you know if HADOOP is indicated to work under "thin" environment
?
>> On Hadoop I will put large workloads and this thin virtualization
> utilizes more resources than exists (shareable environment)
>> that is,
>> if I have a real physical necessity of 500gb for each hadoop host and
> my Thin Virtualization has 2TB on NFS, I can have only 4 virtual machines
> (500GB each), or less.
>>
>> For this case I believe clone virtual machine is the right choice. But
> in my environment it takes 1h30m to build one cloned virtual machine.
>
> if you plan to overcommit then go with thin. The drawback is that if you
> of course hit the physical limit the VMs will run out of space...
> if you plan to allocate 500GB each, consume all of it, never plan to grow
> then go with the "clone"….yes, it's going to take time to write all
that
> stuff. With "thin" you need to do the same amount of writes, but
gradually
> over time while you're allocating it it
>
> hope it helps
>
> Thanks,
> michal
>
>>
>>
>>
>> Am I correct ?
>>
>>
>>
>>
>>
>> On Thu, Apr 17, 2014 at 7:33 AM, Michal Skrivanek <
> michal.skrivanek(a)redhat.com> wrote:
>>
>> On Apr 16, 2014, at 16:41 , Tamer Lima <tamer.americo(a)gmail.com> wrote:
>>
>>>
>>>
>>> Hello,
>>>
>>> I created VMs by two ways :
>>>
>>> 1) on tab virtual machines > new vm > template (centos_65_64bits)
>>> 1.1 configuration : I do not select stateless checkbox
>>> 1.2 this process takes a 1h30 to create each machine.
>>>
>>> 2) on tab pools > new vm > template (centos_65_64bits)
>>> 2.1 default configuration : stateless
>>> 2.2 Here I created 3 virtual machines at once
>>> 2.3 this process takes only one minute
>>>
>>> On the tab virtual machines I can see all virtual machines.
>>> Pooled machines have different icon image
>>> and description is different too:
>>>
>>> machines generated from tab VM are described as clone/dependent
>>> - clone is a phisical copy?
>>> machines generated from tab POOL are described as thin/independent
>>> - thin is a just a reference to template vm ? what is phisical? any
> configuration file?
>>
>> yeah, sort of.
>> just google thin provisioning in general:)
>>
>>
>>>
>>>
>>> In practice, what is the difference between these machines ?
>>>
>>>
>>>
>>>
>>>
http://www.ovirt.org/Features/PrestartedVm
>>> "Today there are 2 types of Vm pools:
>>> • Manual - the Vm is supposed to be manually returned to the
> pool. In practice, this is not really entirely supported.
>>> • Automatic - once the user shuts down the Vm - it returns to
> the pool (stateless)."
>>>
>>> all vm created from pool are stateless ?
>>
>> the automatic pool, yes
>>
>> Thanks,
>> michal
>>
>>>
>>>
>>> thanks
>>>
>>>
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users(a)ovirt.org
>>>
http://lists.ovirt.org/mailman/listinfo/users
>>
>>
>
>
_______________________________________________
Users mailing list
Users(a)ovirt.org
http://lists.ovirt.org/mailman/listinfo/users
--
Mit freundlichen Grüßen / Regards
Sven Kieske
Systemadministrator
Mittwald CM Service GmbH & Co. KG
Königsberger Straße 6
32339 Espelkamp
T: +49-5772-293-100
F: +49-5772-293-333
Geschäftsführer: Robert Meyer
St.Nr.: 331/5721/1033, USt-IdNr.: DE814773217, HRA 6640, AG Bad Oeynhausen
Komplementärin: Robert Meyer Verwaltungs GmbH, HRB 13260, AG Bad Oeynhausen