On Thu, Apr 24, 2014 at 01:04:56PM -0400, Nir Soffer wrote:
----- Original Message -----
> From: "Nir Soffer" <nsoffer(a)redhat.com>
> To: "infra" <infra(a)ovirt.org>
> Sent: Thursday, April 24, 2014 7:47:10 PM
> Subject: cpopen missing again?
>
> Didn't we resolved the issues with cpopen?
>
> nosetests tests.py
> E
> ======================================================================
> ERROR: Failure: ImportError (No module named cpopen)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File "/usr/lib/python2.7/site-packages/nose/loader.py", line 413, in
> loadTestsFromName
> addr.filename, addr.module)
> File "/usr/lib/python2.7/site-packages/nose/importer.py", line 47, in
> importFromPath
> return self.importFromDir(dir_path, fqname)
> File "/usr/lib/python2.7/site-packages/nose/importer.py", line 94, in
> importFromDir
> mod = load_module(part_fqname, fh, filename, desc)
> File
>
"/home/jenkins/workspace/vdsm_master_unit_tests_gerrit/vdsm/lib/zombiereaper/tests.py",
> line 24, in <module>
> from cpopen import CPopen
> ImportError: No module named cpopen
>
> See
http://jenkins.ovirt.org/job/vdsm_master_unit_tests_gerrit/8419/console
Note that triggering again was successful on the same host:
http://jenkins.ovirt.org/job/vdsm_master_unit_tests_gerrit/8420/
Looks like there is a race in the test setup/teardown code, removing and installing
cpopen.
This problem would persist until we mutex jobs that mess with thr rpm
database. Yes: if a job removes an rpm, no other job that adds that rpm
should be scheduled. Yes, even if this means to serialize all jobs on a
single slave.
Eyal - it's better to have a slow CI environment than having a
non-reliable one.