Eric, thanks for taking the time and explaining this in detail. On Mon, 2013-05-13 at 11:09 -0600, Eric Blake wrote: > > There is work underway to switch to an event-based solution - where qemu > periodically asks the guest but without blocking, the guest informs qemu > when it wants, and where libvirt can then query the most recent guest > information (if any) without blocking, which works to remove the hole of > a non-cooperative guest stopping progress. But until libvirt been has > wired up to use the new QMP querying commands for those properties (and > for that matter, until you are using a qemu new enough), there's no way > for libvirt to report a stat that qemu doesn't provide. This seems like a reasonable approach. Virt-manager will then have cater for both cases, when libvirt can provide this info via QMP query commands and when it can't. > https://lists.gnu.org/archive/html/qemu-devel/2013-01/msg04832.html > > I couldn't (quickly) find an open bugzilla against libvirt requesting > that it start using the qom-get command on new-enough qemu, in order to > get guest memory stat reporting back operational. I've also searched RH bugzilla tickets for libvirt (product: Virtualization Tools), but couldn't find a bug report about this. Should I file a new bug report so we can track this? _______________________________________________ virt-tools-list mailing list virt-tools-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/virt-tools-list
