Tempest reports 'missing Worker 1!'

[Ronelle Landy]

Both fully passing and failing Tempest runs show the following at the of the logs:

2017-11-22 16:27:34 | ==============
2017-11-22 16:27:34 | Worker Balance
2017-11-22 16:27:34 | ==============
2017-11-22 16:27:34 | - Worker 0 (286 tests) => 0:38:55.123405
2017-11-22 16:27:34 | - WARNING: missing Worker 1! Race in testr accounting.
2017-11-22 16:27:34 | - Worker 2 (759 tests) => 0:45:58.265308
2017-11-22 16:27:34 | - Worker 3 (308 tests) => 0:27:03.71602

Worker 1 goes missing.

See a full log:
https://logs.rdoproject.org/openstack-periodic/periodic-tripleo-ci-centos-7-ovb-1ctlr_1comp-featureset020-pike/f33faa6/undercloud/home/jenkins/tempest_output.log.txt.gz

[Alex Schultz (alex-schultz)]

I've added os-testr for their visibility. I'm dropping this to a medium since it doesn't appear to fail anything.

[Matthew Treinish (treinish)]

I haven't seen that warning message in a very long time. That is a feature in subunit-trace to warn about a bug we used to see quite frequently in testr: http://git.openstack.org/cgit/openstack/os-testr/tree/os_testr/subunit_trace.py#n303

What version of testr and os-testr are you running with. The most recent release of stestr should be immune from this because it doesn't actually use testr internally anymore (it switched to use stestr). But even if you were using os-testr < 1.0.0 if you're using a new enough version of testr that bug should have been fixed.

[chandan kumar (chkumar246)]

python-os-testr-0.8.2-0.20170811225902.e84d062.el7.centos.noarch and python-testrepository-0.0.18-2.el7.noarch version used https://logs.rdoproject.org/openstack-periodic/periodic-tripleo-ci-centos-7-ovb-1ctlr_1comp-featureset020-pike/f33faa6/rpm-qa.txt.gz grepped from here