coverage task fails on Jenkins
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack DBaaS (Trove) |
Fix Released
|
Undecided
|
Dror Kagan |
Bug Description
it seems that there are (at least) two tests, in the testr unittest framework, that call reddwarf/
this is causing a race condition and these two tests, being run in parallel, will end up failing. the first will actually create the sqlite file and the second will fail complaining that stuff already exists in the sqlite db.
please notice that this doesn't reproduce in a dev environment on a mac.
also notice that this doesn't fail the unittests themselves, since run_tests.py runs first and already set up the sqlite file.
we can't run it in that order (run_tests.py first and testr second) for coverage, since we need to use the -a option for coverage to aggregate the coverage report, and the testr wrapper doesn't expose this option. so we run testr first, and then we run run_tests.py with the -a option.
as a temporary fix, i'll add an execution of run_tests.py in the coverage job before we actually gather coverage, so they don't fail.
however, the proper fix should be to protect the init_db() code to be safe for parallel execution.
Changed in trove: | |
milestone: | none → havana-2 |
status: | Fix Committed → Fix Released |
Changed in trove: | |
milestone: | havana-2 → 2013.2 |
Fix proposed to branch: master /review. openstack. org/24547
Review: https:/