On Mon, May 02, 2016 at 18:35:35 +0200, Michal Privoznik wrote:
On 02.05.2016 16:36, Peter Krempa wrote:
> On Fri, Apr 29, 2016 at 18:43:10 +0200, Michal Privoznik wrote:
>>
https://bugzilla.redhat.com/show_bug.cgi?id=1139766
[...]
> Any failure in qemuProcessReconnect results into libvirtd
killing the
> process. I don't think this failure is criticall enough to allow such
> brutality.
>
> Peter
>
Well, if guest has updated its RTC while we were not running, we would
report spurious time in our XML. Since we are doing something similar
for ejectable disks, i.e. CDROMs too, I presumed it is okay. But I don't
mind changing this to something softer.
As long as it's guaranteed that any target qemu supported by libvirt
does support the interrogation you are going to do it's probably fine.
The problem with your patch is that qom-get was added in qemu 1.1
whereas libvirt supports versions since 0.12
This would effectively reduce compatibility to 1.1 which we should not
do as a side effect of a unrelated change.
Peter