On Wed, Jun 12, 2019 at 09:03:04PM +0200, Paul Gevers wrote: > Hi Bastian, > > [CC adding firstname.lastname@example.org, please drop the bug in the next reply as it > starts to become off-topic there.] > > On 12-06-2019 20:52, Bastian Blank wrote: > > On Wed, Jun 12, 2019 at 08:42:27PM +0200, Paul Gevers wrote: > >> On 12-06-2019 20:01, Bastian Blank wrote: > >>> I'm also not sure if the Debian autopkgtest infrastructure would be able > >>> to do that and build images. The actual testing runs via the Gitlab > >>> CI. > >> You could very probably do it. Depending on how long such a build takes, > > > > One build takes 3 minutes if it runs native and 13 minutes if it runs > > via qemu-user. However it needs awefull amount of network and disk IO. > > I don't believe that should be a problem if that is an *or*. However, > qemu will not work properly, right Antonio?
qemu-user probably works; qemu-system might work, but not with kvm acceleration. > > Does the Debian autopkgtest instance support "needs-root" and > > "breaks-testbed"? > > Yes and yes. The only thing we currently do not support *yet* is > isolation-machine. > > > The image build uses loop devices, hence > > "needs-root", which can't be cleaned up properly, hence > > "breaks-testbed". > > That's no problem at all. mounting loop devices would not work. (these tests probably also need isolation-machine)
Description: PGP signature