Re: OST fails in 002_bootstrap_pytest.py - setup_storage.sh

On Mon, Mar 23, 2020 at 1:25 PM Marcin Sobczyk <msobczyk(a)redhat.com&gt; wrote: > > > On 3/21/20 1:18 AM, Nir Soffer wrote: > > On Fri, Mar 20, 2020 at 9:35 PM Nir Soffer <nsoffer(a)redhat.com&gt; wrote: >> Looks like infrastructure issue setting up storage on engine host. >> >> Here are 2 failing builds with unrelated changes: >> https://jenkins.ovirt.org/job/ovirt-system-tests_manual/6677/ >> https://jenkins.ovirt.org/job/ovirt-system-tests_manual/6678/ > > Rebuilding still fails in setup_storage: > > https://jenkins.ovirt.org/job/ovirt-system-tests_manual/6679/testReport/ > https://jenkins.ovirt.org/job/ovirt-system-tests_manual/6680/testReport/ > >> >> Is this a known issue? >> >> Error Message >> >> AssertionError: setup_storage.sh failed. Exit code is 1 assert 1 == 0 -1 +0 >> >> Stacktrace >> >> prefix = <ovirtlago.prefix.OvirtPrefix object at 0x7f6fd2b998d0> >> >> @pytest.mark.run(order=14) >> def test_configure_storage(prefix): >> engine = prefix.virt_env.engine_vm() >> result = engine.ssh( >> [ >> '/tmp/setup_storage.sh', >> ], >> ) >>> assert result.code == 0, 'setup_storage.sh failed. Exit code is %s' % result.code >> E AssertionError: setup_storage.sh failed. Exit code is 1 >> E assert 1 == 0 >> E -1 >> E +0 >> >> >> The pytest traceback is nice, but in this case it is does not show any useful info. >> >> Since we run a script using ssh, the error message should include the process stdout and stderr >> which probably can explain the failure. > I posted https://gerrit.ovirt.org/#/c/107830/ to improve logging during storage setup. > Unfortunately AFAICS it didn't fail, so I guess we'll have to merge it and wait for a failed job to get some helpful logs. Thanks. It still fails for me with current code: https://jenkins.ovirt.org/job/ovirt-system-tests_manual/6689/testReport/ Same when using current vdsm master.

>> Also I wonder why this code is called as a test (test_configure_storage). This looks like setup >> step so it should run as a fixture. > That's true, but the pytest porting effort was about providing a bare minimum to move away from nose. > Organizing the tests into proper setup/fixtures is a huge task and will be probably implemented > incrementally in the nearest future. Understood