Rally/Updates
Contents
Weekly updates
May 05, 2014
Hello stackers,
due to holidays there hasn't been very much contribution to Rally during the past week. The highlights are as follows:
- Improved CLI output after a benchmark task gets finished. The two tables for atomic actions / overall results now got merged into one very neatly looking table;
- Several important code improvements include getting rid of the undocumented ThreadPool class and also moving some useful functions for benchmarks data processing into a separate module;
- New functionality in Rally for this week is the support for the Ironic client (and potentially also for benchmark scenarios for Ironic) and the introduction of Dummy benchmark scenarios that tend to be very useful for developers while testing their changes.
There is going to be much work this week that includes adding new benchmark scenarios (e.g. for Tempest, Neutron and Ceilometer), further improvements in CLI and graphs and also several bugfixes.
We encourage you to take a look at new patches in Rally pending for review and to help us making Rally better.
Source code for Rally is hosted at GitHub: https://github.com/stackforge/rally
You can track the overall progress in Rally via Stackalytics: http://stackalytics.com/?release=juno&metric=commits&project_type=all&module=rally
Open reviews for Rally: https://review.openstack.org/#/q/status:open+rally,n,z
Stay tuned.
Regards,
The Rally team
April 28, 2014
Hello stackers,
the most important changes during the past week in Rally have been concentrated on creating completely new benchmark scenario classes and adding support for new contexts. To be more precise:
- We've added benchmark scenarios that launch Tempest tests and check the performance of updating quotas for nova/cinder. Both are quite new to Rally and allow users to compose more interesting benchmark cases.
- There is one new context, the "roles" context that allows to change user roles before using those users in benchmark scenarios. Much work has also been done on refactoring the "cleanup" context and helper methods that it uses.
- We've integrated the new oslotest library, wich allowed us to remove much of the "common openstack" code from our repository.
This week, we'll continue our work on expanding the range projects supported by our benchmarks. Benchmark scenarios already pending for review are those for Heat, Ceilometer and also Tempest. The ongoing work also includes several improvents in the overall user experience, e.g. changes in the CLI output format.
We encourage you to take a look at new patches in Rally pending for review and to help us making Rally better.
Source code for Rally is hosted at GitHub: https://github.com/stackforge/rally
You can track the overall progress in Rally via Stackalytics: http://stackalytics.com/?release=juno&metric=commits&project_type=all&module=rally
Open reviews for Rally: https://review.openstack.org/#/q/status:open+rally,n,z
Stay tuned.
Regards,
The Rally team
April 21, 2014
Hello stackers,
here is the report of our recent activities in Rally. They have been quite numerous this week due to the contributions of our constantly growing team and include a couple of changes that touch the end user:
- The scenario runner previously known as continuous has been split to 2 scenario runners named constant and constant_for_duration, thus becoming more "atomic" (previously, the continuous scenario runner was able to run scenarios both for the specified amount of times and for some duration. These two types of behaviour now belong to different runners). "Constant" means that both scenario runners generate a constant load on the cloud;
- Several changes in the input configuration files formats include:
- Renaming the "name" attribute of deployments to "type" in the deployment configuration file. The same change now also applies to server providers;
- Renaming the "active_users" attribute of the constant scenario runners to "concurrency" in the benchmark task configuration files.
- Several improvements in context classes:
- Fixing a couple of issues with the cleaner context: incorrect context initialization and buggy cleanup of Glance resources;
- Improving the user context by making it create the temporary users in concurrent mode by default.
- New benchmark scenarios include those for Keystone tenants and users listing;
This week, we continue our work on new benchmark scenarios for Tempest and quotas, new contexts as well as on many other improvements in Rally.
We encourage you to take a look at new patches in Rally pending for review and to help us making Rally better.
Source code for Rally is hosted at GitHub: https://github.com/stackforge/rally
You can track the overall progress in Rally via Stackalytics: http://stackalytics.com/?release=juno&metric=commits&project_type=all&module=rally
Open reviews for Rally: https://review.openstack.org/#/q/status:open+rally,n,z
Stay tuned.
Regards,
The Rally team
April 14, 2014
Hello stackers,
let us list our main contributions to Rally during the past week:
- The Bechmark context manager class, which mostly finishes the work on contexts integration in Rally; this manager allows Rally to use only those contexts that are really needed for the current benchmark scenario, thus greatly reducing the runtime of the preparation step, manages the order of entering those contexts, adds logging for contexts and does a couple of other important things;
- Improvement of the CLI for the Tempest verifier: we've added the functionality of saving verification results into a database and also a range of new commands to handle these results, such as rally verify list, rally verify detailed and others;
- Validation for required parameters in benchmark scenarios has now been implemented through a new validator, added, as usual, as a decorator to the corresponding benchmark scenarios;
This week, our work continues with several new benchmark scenarios (including those for Tempest), new validators and also new functionality in the Rally core, e.g. the ability to use already existing tenants/users for benchmarking, implemented now through contexts.
We encourage you to take a look at new patches in Rally pending for review and to help us making Rally better.
Source code for Rally is hosted at GitHub: https://github.com/stackforge/rally
You can track the overall progress in Rally via Stackalytics: http://stackalytics.com/?release=icehouse&metric=commits&project_type=all&module=rally
Open reviews for Rally: https://review.openstack.org/#/q/status:open+rally,n,z
Stay tuned.
Regards,
The Rally team
April 7, 2014
Hello stackers,
our contributions this week have been pretty diverse and touched different parts of Rally. Let us name a few that seem to be the most interesting ones:
- Nova and Cinder quotas management through a new context class. Another example of how useful and easy to use contexts are in Rally;
- Concurrent users creation and deletion in the user context. This change greatly reduces the runtime of this benchmarking step;
- New benchmark scenarios for Keystone and Glance;
- Nice enhancement of the auto-generated report html page through adding histograms for runtimes there.
There are also other parts of Rally which are in progress now and will hopefully have their patches merged the next week, e.g. the work on Tempest integration or further context classes implementation.
We encourage you to take a look at new patches in Rally pending for review and to help us making Rally better.
Source code for Rally is hosted at GitHub: https://github.com/stackforge/rally
You can track the overall progress in Rally via Stackalytics: http://stackalytics.com/?release=icehouse&metric=commits&project_type=all&module=rally
Open reviews for Rally: https://review.openstack.org/#/q/status:open+rally,n,z
Stay tuned.
Regards,
The Rally team