[VDSM] ipwrapperTests.TestDrvinfo failing on f21 slave, pass on f20 slave

[Nir Soffer]

----- Original Message -----
> From: "Dan Kenigsberg" <danken at redhat.com>
> To: "Nir Soffer" <nsoffer at redhat.com>
> Cc: "infra" <infra at ovirt.org>, "devel" <devel at ovirt.org>, "Ido Barkan" <ibarkan at redhat.com>
> Sent: Thursday, May 7, 2015 12:32:03 PM
> Subject: Re: [VDSM] ipwrapperTests.TestDrvinfo failing on f21 slave, pass on f20 slave
> 
> On Wed, May 06, 2015 at 05:28:14PM -0400, Nir Soffer wrote:
> > Hi infra,
> > 
> > These tests fail on Fedora 21 slave, and pass on Fedora 20 slave.
> > http://jenkins.ovirt.org/job/vdsm_master_unit-tests_created/1888/console -
> > FAIL
> > http://jenkins.ovirt.org/job/vdsm_master_unit-tests_created/1908/console -
> > PASS
> > 
> > Infra: please check the Fedora 21 slave, maybe this is configuration issue?
> > Ido: please check if the tests are valid
> > 
> > Generally these tests are not unitests, and should not run for validating
> > unrelated patches. They should move to the functional tests.
> 
> Could you explain your assertions?
> 
> This tests do not require a functioning vdsm. They don't test vdsm's
> API, so they should not end up in the "functional" directory.
> 
> The ipwrapperTests tests the ipwrapper unit of vdsm. But in order to do
> so, they interact with the host system, and attempt to create a bridge.
> When this setup fails for some reason (I do not know why `brctl
> addbr vdsmtest-O8JmI` failed) the test ends up in an ERROR state.
> 
> If we fail to fix this condition ASAP, we catch this failure and skip
> the test as broken. But I don't think we should thrown it out of the
> test suite altogether.

This requires functioning brctl, so this is integration test. This is true
for many other tests that should move to the missing integration tests suite.

Nir