Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Fri, May 08, 2015 at 03:11:16PM +0200, Philipp Marek wrote: > > configuration for live migration of QEMU/libvirt resources. I evaluated > > possible use of OCFS2, but ended up just using DRBD dual primary mode, which is > > just as well, as pcmkr support for live migration is insufficient for our > > purpose: http://bugs.clusterlabs.org/show_bug.cgi?id=5055 > Ah, okay. > > Yes, for the few cases where online migration is need, our admins typically > put the cluster in maintenance-mode, migrate the (small number of) VMs > "manually", and clean up again. Phillip stated that the drbd staff do "manual" (not using pacemaker) live migration of their VMs, (temporarily) using drbd dual primary mode. I'm curious if there's a (published) document about this procedure? Or if you do anything more involved than this: crm -w configure property maintenance-mode=true drbdadm net-options --allow-two-primaries all drbdadm disk-options --fencing=resource-and-stonith all # per node drbdadm primary all virsh migrate --live --verbose database qemu://bg4/system drbdadm secondary all drbdadm adjust all # per node crm -w resource reprobe crm -w configure property maintenance-mode=false Thanks, Justin