description of problem: vm floppy device iotune will be cleared after update-device but can get it via dumpxml Version-Release number of selected component (if applicable): libvirt-1.2.8-11.el7.x86_64 How reproducible: 100% Steps to Reproduce: 1.prepare a guest have floppy device: <disk type='file' device='floppy'> <driver name='qemu' type='qcow2' cache='none'/> <source file='/var/lib/libvirt/images/test6.img'/> <target dev='fda' bus='fdc'/> <shareable/> <address type='drive' controller='0' bus='0' target='0' unit='0'/> </disk> 2.start it and set iotune in this device: # virsh blkdeviotune test4 fda 123 --live # virsh blkdeviotune test4 fda total_bytes_sec: 123 read_bytes_sec : 0 write_bytes_sec: 0 total_iops_sec : 0 read_iops_sec : 0 write_iops_sec : 0 3.prepare a xml like this (for update-device): # cat floppydisk.xml <disk type='file' device='floppy'> <driver name='qemu' type='qcow2' cache='none'/> <source file='/var/lib/libvirt/images/test6.img'/> <target dev='fda' bus='fdc'/> <shareable/> <iotune> <total_bytes_sec>1234</total_bytes_sec> </iotune> </disk> 4.do update-device # virsh update-device test4 floppydisk.xml Device updated successfully 5.cannot see iotune via blkdeviotune: # virsh blkdeviotune test4 fda total_bytes_sec: 0 read_bytes_sec : 0 write_bytes_sec: 0 total_iops_sec : 0 read_iops_sec : 0 write_iops_sec : 0 6.can see it via dumpxml(the old settings) # virsh dumpxml test4 <disk type='file' device='floppy'> <driver name='qemu' type='qcow2' cache='none'/> <source file='/var/lib/libvirt/images/test6.img'/> <backingStore/> <target dev='fda' bus='fdc'/> <iotune> <total_bytes_sec>123</total_bytes_sec> </iotune> <shareable/> <alias name='fdc0-0-0'/> <address type='drive' controller='0' bus='0' target='0' unit='0'/> </disk> Actual results: vm floppy device iotune will disappear but can see it via dumpxml after update-device Expected results: dumpxml and blkdeviotune get the same result after use update-device Additional info:
I just tried to reproduce, and the iotune values reported in XML and and via blkdeviotune output match, so I guess this is fixed