Description of problem: when setting io mode to default in virt-manager for block device,should automatic attach io=native to guest Version-Release number of selected component (if applicable): virt-manager-0.9.0-5.el6 python-virtinst-0.600.0-2.el6 libvirt-0.9.4-2.el6 qemu-kvm-0.12.1.2-2.177.el6 How reproducible: always Steps to Reproduce: 1.install guest with a block device. #virsh dumpxml demo ..... <disk type='block' device='disk'> <driver name='qemu' type='raw' cache='none' io='native'/> <source dev='/dev/sda2'/> <target dev='hdb' bus='ide'/> <alias name='ide0-0-1'/> <address type='drive' controller='0' bus='0' unit='1'/> </disk> ..... the io mode set to native automatic. 2.shutdown the guest ,in virt-manager ,switch to guest detail page 3.change io mode to default and apply 4.start the guest Actual results: in guest xml,io='native' removed, #ps -ef | grep qemu-kvm aio=native removed from block device Expected results: the qemu default io mode is threads,actually block device default mode is native, so should add io='native' when setting default mode for block device Additional info:
Hmm, we have usually had 'default' mean 'hypervisor default' so the behavior is consistent, but I agree it's a bit misleading.
Not urgent, and given reduced capacity for virt-manager/virtinst, just moving this to the upstream tracker.
Actually, reconsidering, this is the best we can do, short of having libvirt and qemu report default capabilities to us. Trying to hardcode what the default is just causes maintenance pain, especially in this case when it isn't having any effect on actual configuration. Closing as WONTFIX.