Hello,
Can't really help you with you with nova, but using plain libvirt-1.1.1
and qemu-1.5.3 live migration of rbd-backed VMs is (almost*) instant on
the client side. We have rbd write-back cache enabled everywhere and
have no problem at all.
-K.
*There is about a 1-2 second hitch at worse, I suppose the switches need
that much to notice the change.
On 04/16/2015 06:45 AM, Yuming Ma (yumima) wrote:
The issue is reproducible in svl-3 with rbd cache set to false.
On the 5th ping-pong, the instance experienced ping drops and did not
recover for 20+ minutes:
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
