Hi again, On 10 April 2014 16:17, Haomai Wang <haomaiw...@gmail.com> wrote:
> I think you need to bind osd to specified cores and bind qemu-kvm to > other cores. Perhaps, but that would just be an optimisation. More likely we'd just bind the OSDs to the socket that is handling the SAS/RAID controller. Memory size is another factor need to take care of. This is one that worries me a bit with respect to just running OSDs unbounded/un-containered on the hosts, as there seem to have been a few incidences of OSDs running rampant in memory previously. I think perhaps a cgroup specifying cpu and memory constraints would be the way to go, but I wonder if anyone has done this? > If > your vm's root disk uses local file, the IO problem maybe > intractability > No, I'm suggesting we would run Ceph for ephemeral storage as an alternative to qcow/raw vdisk files on a local filesystem (what we do now). On Thu, Apr 10, 2014 at 12:47 PM, Blair Bethwaite > > Is anyone else doing this successfully, is there anything else to > consider? > Nobody? Don't think we want to go down a road untraveled. -- Cheers, ~Blairo
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
