On Fri, Oct 10, 2014 at 4:47 PM, Richard W.M. Jones <rjones(a)redhat.com&gt; wrote: > On Fri, Oct 10, 2014 at 04:26:20PM +0800, Zhi Yong Wu wrote: >> You can't make sure that guest never fail or panic forever, right? > > Do you mean the guest or the libguestfs appliance? > >> That is, you will have to maintain that guest, not customers' guests. >> This risk is beyond thought. The maintainance cost will be very >> expensive in our super large scale cloud environment. >> >> We are trying not to start a appliance guest when issuing a libguestfs >> command, such as virt-resize. > > There may be some confusion here. Virt-resize doesn't run the guest. When i issue virt-resize, i exactly saw one guest is started, indisk and outdisk are attached to this guest.... Do you mean that in the latest upstream, virt-resize doesn't run the guest?

> TBH I'm not exactly sure what your concerns are. We are afraid that if a guest is started when vrit-xxx command is issued, it will bring some unexpected risks and take too much time to complete this command.