Hi,
When you'll press 'ok', a snapshot of that disk's image chain (its Base volume) is created in the source storage domain, and the entire image chain is replicated in the destination storage domain.
It doesn't effect the original disk, and meanwhile, the VM keep "act normally":

- If the VM's actions only deal with reading the disk, creating its snapshot should be a simpler task.
- In case the VM writes data to the disk, a new snapshot volume which contains only the changes (for simplifying, you can think the way "git diff" works) is being created on both source and destination targets.
​While the base volume is being copied from the source target to the destination target​, those snapshot volumes (for disk's changes) get synchronized.

When the disk's images on both source and destination targets are identical, the VM points to the 'new' image and deletes the old pointer.  
I.e. the disk was successfully moved.

In case of any failure during the migration, since targeted destination contains only a snapshot of the original disk's image, the original image isn't being effected,
So there won't be a data lose and operation just fails.
Also, busy network can affect the migration's duration.   


Regards,
Shani Leviim

On Sun, Nov 26, 2017 at 5:38 PM, Arman Khalatyan <arm2arm@gmail.com> wrote:
hi Sahni,
thanks for the details.
Looks like the live storage migration might fail on the heavy loaded virtual machines. I just tried to move from nfs to iscsi storage(most of the cases they moved w/o error), the message on the move dialog warns us  "! moving following disks when VM is running", if we press "ok ! do it" what are the consequences? it is not explained in the docs. 
thank you beforehand
Arman.


Am 26.11.2017 2:12 nachm. schrieb "Shani Leviim" <sleviim@redhat.com>:
Hi Arman,
VM's migration and disks migration are two different things:

- Live storage migration:
A VM's disk can be moved to another storage domain while the VM is running, by copying the disk's structure the destination domain.
The hard part of live storage migration is moving the active layer volumes from one domain to another, while the VM is writing to those volumes.
By using a replication operation, the data is written to both source and destination volumes.
When both volumes contain the same data, the block job operation can be aborted, pivoting to the new disk.


- Live migration:
Provides the ability to move a running virtual machine between physical hosts with no interruption to service.
The virtual machine remains powered on and user applications continue to run while the virtual machine is relocated to a new physical host.
A running virtual machine can be live migrated to any host within its designated host cluster.
Live migration of virtual machines does not cause any service interruption.


Hope it helps!

Regards,
Shani Leviim

On Fri, Nov 24, 2017 at 11:54 AM, Arman Khalatyan <arm2arm@gmail.com> wrote:
hi,
I have some test enviroment with ovirt
"4.2.0-0.0.master.20171114071105.gitdfdc401.el7.centos"
2hosts+2NFS-domains

During the multiple disk movement between the domains I am getting this warning:
Moving disks while the VMs are running.(this is not so scary red as in
4.1.x :) )

What kind of problems can happen during the movement?

Thanks,
Arman.
_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users