[Qemu-discuss] raw block -> RBD live migration

[Nikola Ciprich]

Hello qemu users,

I'd like to ask about nonshared migration..

I have a server with virtual machines running on raw block devices..
I'd like to migrate those to ceph / RBD storage without downtime.

I've tested following procedure (I'm using qemu-kvm + libvirt based management)

1) create RBD devices of the same size
2) create new XML domain definition file with storage changed to new RBD 
devices (vm-new.xml)
3) migrate vm using libvirt:
migrate --verbose --xml /tmp/vm-new.xml --copy-storage-all --live vmlbx50 
qemu+tcp://lbxphav1a/system

this seemed to do the trick, I've tested it for both linux and windows machines.

however, one windows 2k12 machine slowed down badly after migration, and after 
reboot,
chkdisk showed lots of corrupted filesystem indexes (i'm not windows guru, so
hopefully I'm not confusing this much). After fixing those, everything started
working OK again, but I'm now concerned whether this procedure is really safe..

is there enybody who could confirm or comment on this?
I'm using qemu-2.2.1 and libvirt 1.2.6 on source machine,
qemu-2.3.0 and libvirt 1.2.15 on target box..

thanks a lot in advance!

BR

nik

-- 
-------------------------------------
Ing. Nikola CIPRICH
LinuxBox.cz, s.r.o.
28.rijna 168, 709 00 Ostrava

tel.:   +420 591 166 214
fax:    +420 596 621 273
mobil:  +420 777 093 799
www.linuxbox.cz

mobil servis: +420 737 238 656
email servis: address@hidden
-------------------------------------

pgpRkhXdBNLYL.pgp
Description: PGP signature