On 11/18/2014 10:17 AM, Victor Sergeyev wrote:
Matt, As for race in Nova - it caused by deprecated is_backend_avail() function, which calls _ensure_backenv_available() method, which creates a SQLAlchemy engine and opens a test connection, but doesn't call engine.dispose(). Depending on Python interpreter version, this connection may remain open for some time. So there are such ways to fix Nova: - wait for oslo.db 1.1.1 which will include fix for this method - see patch [1] - remove is_backend_avail() helper usage in Nova - patch [2] refactor Nova opportunistic DB tests and remove that method. [1] https://review.openstack.org/#/c/135293/ [2] https://review.openstack.org/#/c/103920/ On Tue, Nov 18, 2014 at 5:22 AM, Matt Riedemann <[email protected] <mailto:[email protected]>> wrote: On 11/17/2014 9:36 AM, Victor Sergeyev wrote: Hello All! Oslo team is pleased to announce the new release of Oslo database handling library - oslo.db 1.1.0 List of changes: $ git log --oneline --no-merges 1.0.2..master 1b0c2b1 Imported Translations from Transifex 9aa02f4 Updated from global requirements 766ff5e Activate pep8 check that _ is imported f99e1b5 Assert exceptions based on API, not string messages 490f644 Updated from global requirements 8bb12c0 Updated from global requirements 4e19870 Reorganize DbTestCase to use provisioning completely 2a6dbcd Set utf8 encoding for mysql and postgresql 1b41056 ModelsMigrationsSync: Add check for foreign keys 8fb696e Updated from global requirements ba4a881 Remove extraneous vim editor configuration comments 33011a5 Remove utils.drop_unique_constraint() 64f6062 Improve error reporting for backend import failures 01a54cc Ensure create_engine() retries the initial connection test 26ec2fc Imported Translations from Transifex 9129545 Use fixture from oslo.config instead of oslo-incubator 2285310 Move begin ping listener to a connect listener 7f9f4f1 Create a nested helper function that will work on py3.x b42d8f1 Imported Translations from Transifex 4fa3350 Start adding a environment for py34/py33 b09ee9a Explicitly depend on six in requirements file 7a3e091 Unwrap DialectFunctionDispatcher from itself. 0928d73 Updated from global requirements 696f3c1 Use six.wraps instead of functools.wraps 8fac4c7 Update help string to use database fc8eb62 Use __qualname__ if we can 6a664b9 Add description for test_models_sync function 8bc1fb7 Use the six provided iterator mix-in 436dfdc ModelsMigrationsSync:add correct server_default check for Enum 2075074 Add history/changelog to docs c9e5fdf Add run_cross_tests.sh script Thanks Andreas Jaeger, Ann Kamyshnikova, Christian Berendt, Davanum Srinivas, Doug Hellmann, Ihar Hrachyshka, James Carey, Joshua Harlow, Mike Bayer, Oleksii Chuprykov, Roman Podoliaka for contributing to this release. Please report issues to the bug tracker: https://bugs.launchpad.net/__oslo.db <https://bugs.launchpad.net/oslo.db> _________________________________________________ OpenStack-dev mailing list [email protected].__org <mailto:[email protected]> http://lists.openstack.org/__cgi-bin/mailman/listinfo/__openstack-dev <http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev> And...the nova postgresql opportunistic DB tests are failing quite frequently due to some race introduced by the new library version [1]. [1] https://bugs.launchpad.net/__oslo.db/+bug/1393633 <https://bugs.launchpad.net/oslo.db/+bug/1393633> -- Thanks, Matt Riedemann _________________________________________________ OpenStack-dev mailing list [email protected].__org <mailto:[email protected]> http://lists.openstack.org/__cgi-bin/mailman/listinfo/__openstack-dev <http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev> _______________________________________________ OpenStack-dev mailing list [email protected] http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Could https://review.openstack.org/#/c/135293/ have any impact on unit tests that hit a backing sqlite database, even non-db migration unit tests?
The reason I ask is running the full nova tox -re py27 job on my 4 VCPUs Ubuntu Trusty (python 2.7.6) is taking noticeably longer recently. I don't see any recent changes to the nova.tests.integrated tree that might have caused a problem, that doesn't mean it might not be something else that recently changed in nova, just grasping at straws right now (for all I know I've got some rogue process on my system eating CPU and RAM too).
-- Thanks, Matt Riedemann _______________________________________________ OpenStack-dev mailing list [email protected] http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
