For the record,
here the patch send to the mailing
http://pve.proxmox.com/pipermail/pve-devel/2014-November/013080.html

They are 2way to do it,

1) old way 
----------
use qmp migrate with blk=>  option.

This copy disk to target vm, with exactly same path.
It's work great if you have only local disks, and target have same disk/path.
But If you have a shared disk + local disk it don't work.

The datas are transferred inside the same thread than live migration.

All datas of all disk are transferred, and the delta of writes are done before 
finish the livemigration.


This is what the patch from the mailing is doing. (Need some cleanup, but 
pretty easy to implement)


2) new way
----------
using block-mirror and nbd

a)starting target vm and expose disk with nbd
b) do block-mirror to target vm to nbd
c) at the end of copy block-mirror, switch sourcevm disk to target nbd
d) do the live-migration

Advantage is that we can choose which disk we want to mirror.
we can have some local disk and some shared disk, and only copy local disk.
Also target disk can be a different storage/path.

The problem is that if you mirror multiple disk and 
for example:
you have finished to mirror first disk (so source vm is writing to target nbd)
and the mirror is failing on 2nd disk.
I don't known how to rollback.
Because you can have already new write on first target disk, and also write on 
source 2nd disk


So it's more flexible but more complex to implement




----- Mail original -----
De: "Wolfgang Link" <wolfg...@linksystems.org>
À: "Lindsay Mathieson" <lindsay.mathie...@gmail.com>
Cc: "pve-devel" <pve-devel@pve.proxmox.com>
Envoyé: Mercredi 29 Avril 2015 06:43:51
Objet: Re: [pve-devel] [PATCH_V2] implement offline migration on zfs



Hi, 
As fare I know is hyper-v shared nothing a live mig. 
Or Migration is a offline only migration. 

About the speed, this is at the moment limited by openssh. 
Am 29.04.2015 00:50 schrieb "Lindsay Mathieson" < lindsay.mathie...@gmail.com 
>: 


Just as a matter of curiosity - this is Live offline Migration? like Hyper-V's 
"Shared Nothing" live migration? 

Also, is it any faster than std storage migration? 

-- 
Lindsay Mathieson 
_______________________________________________ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 




_______________________________________________ 
pve-devel mailing list 
pve-devel@pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 
_______________________________________________
pve-devel mailing list
pve-devel@pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Reply via email to