On 05/22/2014 03:20 PM, Eric Blake wrote:
On 05/22/2014 07:47 AM, Peter Krempa wrote:
> Refresh the disk backing chains when reconnecting to a qemu process
> after daemon restart. There are a few internal fields that don't get
> refreshed from the XML. Until we are able to do that, let's reload all
> the metadata by the backing chain crawler.
Not necessarily ideal. Reading metadata from a file that is also
simultaneously opened read-write by qemu might, in super-rare
circumstances, hit a race where qemu is in the middle of updating
metadata due to the completion of some job that was started before the
libvirtd restart.
For that matter, all of our other places where we reread the backing
chain while qemu is still in operation are also suspect, so your patch
isn't making it any worse :)
ACK as-is.
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org