[Users] from irc

[Itamar Heim]

On 11/15/2013 01:15 PM, Bob Doolittle wrote:
>
> On 11/15/2013 12:36 PM, Gianluca Cecchi wrote:
>> Actually AFAIK in rhev 3.2 (and in oVirt 3.2) pools can be automatic
>> or manual
>> (https://access.redhat.com/site/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.2/html/Administration_Guide/sect-Virtual_Machine_Pool_Tasks.html)
>>
>> "
>> In the Pool tab, select one of the following pool types:
>> Manual - The administrator is responsible for explicitly returning the
>> virtual machine to the pool. The virtual machine reverts to the
>> original base image after the administrator returns it to the pool.
>> Automatic - When the virtual machine is shut down, it automatically
>> reverts to its base image and is returned to the virtual machine pool.
>> "
>>
>> IN RHEV 2.x there was also a third option that was time based and that
>> in my opinion was interesting. Donna why it was removed.
>>
>> In my opinion there is sort of mislead between what is above and the
>> introduction of pools inside the guide because it is stated this way:
>> (https://access.redhat.com/site/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.2/html/Administration_Guide/chap-Pools.html)
>>
>> "
>> ...
>> Virtual machines in pools are stateless, data is not persistent across
>> reboots. Virtual machines in a pool are started when there is a user
>> request, and shut down when the user is finished.
>> ..."
>>
>> So if one doesn't go ahead he/she thinks only stateless are allowed.....
>>
>
> That is very interesting. After reading the RHEV docs I had wondered
> about this myself - i.e. how does one create/manage a stateful model
> where people always return to VMs in the same state they were left? I
> thought perhaps you had to detach it from the Pool for this, but haven't
> had time to experiment with it yet.
>
> With VMware there is a stateful pool model where the user-to-machine
> binding remains in effect, but the machine can be rebooted, powered off,
> etc. Is this what's implied by the "Manual" model?
>

yes. until the admin returns it to the pool, when it looses the snapshot.
allows to easily create a pool of VMs, assign to a group, then each gets 
their own VM (even "forever"), instead of creating/assigning one by one.