Rally v0.3.2


Commits 55
Dev cycle 25 days
Release date 3/14/2016


This release, as well as all previous ones, includes a lot of internal and external changes. Most important of them are listed below.

CLI changes

  • Warning

    [Modified] Option '--tempest-config' for 'rally verify reinstall' command was deprecated for removal.

  • Warning

    [Removed] Option --system-wide-install was removed from rally verify commands in favor of --system-wide option.

  • Warning

    [Modified] Step of installation of Tempest during execution of the rally verify start command was deprecated and will be removed in the future. Please use rally verify install instead.

  • Rework commands.task.TaskCommands.detailed. Now output of the command contains the same results as in HTML report.

Rally Verify

  • Re-run failed Tempest tests

    Add the ability to re-run the tempest tests that failed in the last test execution. Sometimes Tempest tests fail due to a special temporary condition in the environment, in such cases it is very useful to be able to re-execute those tests.

    Running the following command will re-run all the test that failed during the last test execution regardless of what test suite was run.

    rally verify start --failing


  • Scenarios:
  • [updated] Fix flavor for cloudera manager

    Cloudera manager need master-node flavor

  • [added] Expand Nova API benchmark in Rally

    Add support for listing nova hosts, agents, availability-zones and aggregates.

  • [updated] Make sure VolumeGenerator uses the api version info while cleanup

  • Designate V2 - Add recordset scenarios

    Add create_and_(list|delete)_recordset scenarios Remove the test also that checks the allowed methods, this is in order for us to be able to have a private method _walk_pages that will do fetching of pages for us vs attempting to fetch 1 giant list at once.

  • unify *_kwargs name in scenarios

    When running a scenario, kwargs is used as default key-word arguments. But in some scenarios, there are more and one services being called, and we use xxx_kwargs for this case.

    However, some xxx_kwargs are not unified for same usage[0]. Unifying these could avoid misleading for end users. Another improvement is to add xxx_kwargs with empty settings for scenario config files.

    [0] http://paste.openstack.org/show/489505/

  • Warning

    Deprecated arguments 'script' and 'interpreter' were removed in favor of 'command' argument.

    VM task scenarios executes a script with a interpreter provided through a formatted argument called 'command' which expects the remote_path or local_path of the script and optionally an interpreter with which the script has to be executed.


  • Avoid using len(x) to check if x is empty

    This cases are using len() to check if collection has items. As collections have a boolean representation too, directly check for true / false. And fix the wrong mock in its unit test.

  • Fix install_rally.sh to get it to work on MacOSX

    On MacOSX, mktemp requires being passed a template. This change modifies the calls to mktemp to explicitly pass a template so that the code works on both MacOSX and linux.

  • Use new-style Python classes

    There are some classes in the code that didn't inherited from nothing and this is an old-style classes. A "New Class" is the recommended way to create a class in modern Python.A "New Class" should always inherit from object or another new-style class.

    Hacking rule added as well.

  • Make Rally cope with unversioned keystone URL

    With the change, the client version that's returned is now determined by the keystoneclient library itself based on whether you supply a URL with a version in it or not.

  • Fix rally-mos job to work with mos-8.0

    Also remove hardcoded values for some other jobs.

  • Add name() to ResourceManager

    This will allow us to perform cleanup based on the name.

  • Add task_id argument to name_matches_object

    This will be used to ensure that we are only deleting resources for a particular Rally task.

  • Extend api.Task.get_detailed

    Extend api.Task.get_detailed with ability to return task data as dict with extended results.

Bug fixes

The most critical fixed bugs are:

  • #1547624: Wrong configuration for baremetal(ironic) tempest tests

  • #1536800: openrc values are not quoted

    The openrc file created after rally deployment --fromenv did not quote the values for environment variables that will be exported.

  • #1509027: Heat delete_stack never exits if status is DELETE_FAILED

  • #1540545: Refactored atomic action in authenticate scenario

  • #1469897: Incompatible with Keystone v3 argument in service create scenario

  • #1550262: Different results in rally task detailed, rally task report and rally task status commands.

  • #1553024: Backward incompatible change in neutronclient(release 4.1.0) broke Tempest config generation to support latest neutronclient.


  • Add documentation for DB migration
  • Make documentation for output plugins
  • Documentation tox fix

    Added information about debugging unit test with tox. Replace 3 references to py26 with py34 to reflect current rally tox configuration.

  • Change structure of rally plugin and plugin references page

  • Update the scenario development, runner and context sections

  • The design of Rally Plugins Reference page was improved


To Everybody!