Today we're pleased to announce Rally 0.2.1.
- ... upgrade:
pip3 install --upgrade esrally
- ... install:
pip3 install esrally
Please follow the installation instructions for a first time install though.
Rally now has a tournament mode which allows you to compare performance across races
esrally list races to find previously recorded races. Then compare races by their race timestamp, e.g.
esrally compare --baseline=20160502T190127Z --contender=20160502T191011Z. See the docs on tournament mode for more details.
Be aware that the tournament mode is not yet ideal as there is natural variance across races. Therefore, we will run multiple rounds in each race with the next release to account for this variance.
User Defined Tags
Rally can now add one user-defined tag to each metrics record captured during a race. This can be used to split data in Kibana or to memorize what a race was about when running
esrally list races. Just specify the tag on the command line with
On Linux, Rally can now use the perf profiler to gather CPU performance counters. Data is written into a separate log file after each race. See the docs for more details.
We have added a track (benchmark data file) consisting of half a million academic papers (under the CC license) to benchmark a fulltext scenario. You can run it with
esrally --track=pmc but keep in mind the benchmark data are 6GB (compressed). In total you need roughly 70GB of free disk space to run this track.
We also added a track to benchmark geo queries in Elasticsearch which you can run with
esrally --track=geopoint. This track needs about 7GB of free disk space.
- Kudos to Lyndon Swan for adding support for additional compression formats in Rally: https://github.com/elastic/rally/pull/88
If you have questions or feedback, please just post in the Elasticsearch forum.