Rally v0.2.0

Information

Commits 48
Bug fixes 6*
Dev cycle 19 days
Release date 1/11/2015

Details

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

Warning

Release 0.2.0 doesn't support python 26

Deprecations

  • Option --system-wide-install for rally verify start was deprecated in favor of --system-wide

  • rally show commands were deprecated because of 3 reasons:
    • It blocks us to make Rally generic testing tool
    • It complicates work on Rally as a Service
    • You can always use standard OpenStack clients to do the same

Rally Verify

  • Add "xfail" mechanism for Tempest tests.

    This mechanism allows us to list some tests, that are expected to fail, in a YAML file and these tests will have "xfail" status instead of "fail".

    Use new argument "--xfails-file" of rally verify start command.

Rally Task

  • --out argument of rally task report is optional now

    If you don't specify --out <file> it will just print the resulting report

  • Better scenario output support

    As far as you know each scenario plugin are able to return data as a dict. This dict contained set of key-values {<name>: <float>} where each name was line on graph and each number was one of point. Each scenario run adds a single point for each line on that graph.

    This allows to add extra data to the Rally and see how some values were changed over time. However, in case when Rally was used to execute some other tool and collect it's data this was useless.

    To address this Scenario.add_output(additive, complete) was introduced:

    Now it is possible to generate as many as you need graphs by calling this method multiple times. There are two types of graph additive and complete. Additive is the same as legacy concept of output data which is generated from results of all iterations, complete are used when you would like to return whole chart from each iteration.

    HTML report has proper sub-tabs Aggregated and Per iteration inside Scenario Data tab.

    Here is a simple example how output can be added in any scenario plugin:

    # This represents a single X point in result StackedArea.
    # Values from other X points are taken from other iterations.
    self.add_output(additive={"title": "How do A and B changes",
                              "description": ("Trend for A and B "
                                              "during the scenario run"),
                              "chart_plugin": "StackedArea",
                              "data": [["foo", 42], ["bar", 24]]})
    # This is a complete Pie chart that belongs to this concrete iteration
    self.add_output(
        complete={"title": "",
                  "description": ("Complete results for Foo and Bar "
                                  "from this iteration"),
                  "chart_plugin": "Pie",
                  "data": [["foo", 42], ["bar", 24]]})
    

Rally Certification

None.

Specs & Feature Requests

Plugins

  • Scenarios:
  • [new] DesignateBasic.create_and_update_domain

  • [improved] CinderVolumes.create_and_attach_volume

    Warning

    Use "create_vm_params" dict argument instead of **kwargs for instance parameters.

  • Context:
  • [improved] images

    Warning

    The min_ram and min_disk arguments in favor of image_args, which lets the user specify any image creation keyword arguments they want.

Bug fixes

6 bugs were fixed:

  • #1522935: CinderVolumes.create_and_attach_volume does not accept additional args for create_volume
  • #1530770: "rally verify" fails with error 'TempestResourcesContext' object has no attribute 'generate_random_name'
  • #1530075: cirros_img_url in rally.conf doesn't take effective in verification tempest
  • #1517839: Make CONF.set_override with parameter enforce_type=True by default
  • #1489059: "db type could not be determined" running py34
  • #1262123: Horizon is unreachable outside VM when we are using DevStack + OpenStack

Documentation

None.

Thanks

2 Everybody!