Jeff > I live migrate the VM's to the node that is staying up, then back again when the reboot is complete. 2 questunes here , - what happen if the reboot is at night(self reboot of the server) iv seen it happen and nobody there to do the migration ? the all thing breaks ? - what do you mean migrating ? as i understand kvm migration is move images from a server to another with the same storage so the storage is the same anyway . also migration will need to use the gluster mount point anyway so we stay with the running server only while the other still offline The all things looks very uncomftable or am i wrong ? I thought maybbe there is a way to work this without bothering KVM etc.. :) and just let it work , kind of laisy with this things :) 2009/9/1, Jeff Evans <jeffe at tricab.com>: > > Hi, > >> But what about if the file is a big image file of xen or kvm > > I run replicate across two nodes, upon which I host xen images. > I often reboot one of the nodes while the xen machines are running. > > I live migrate the VM's to the node that is staying up, then back > again when the reboot is complete. > > When migrating back, gluster re-syncs the image no problems. > > The re-sync does cause a xen migration or a xen create to take longer, > of course, but it all works as expected. > > Self-healing works fine, just as long as you never access the > underlying FS directly. > > Jeff. > > >
