Rally 0.3.0 released

(Daniel Mitterdorfer) #1

Today we're pleased to announce Rally 0.3.0.

How to...?

  • ... upgrade: pip3 install --upgrade esrally
  • ... install: pip3 install esrally

Note: Depending on your system setup you probably need to prepend these commands with sudo.

Please follow the installation instructions for a first time install though.

Release Highlights

Simplified initial configuration of Rally

Rally autodetects available third-party software now and sets sensible defaults. A separate metric store is not required anymore (although it's still possible to use one for advanced use cases).

Tracks are external to Rally

A "track" is a description of a benchmark in Rally speak. Previously they have been included with Rally as Python files. They live now in a separate repository at https://github.com/elastic/rally-tracks. Doing this enabled the next change:

Ability to benchmark Elasticsearch 1.7, 2.x and 5

Previously, Rally was only able to benchmark Elasticsearch 5 but with the new track repository concept you can now benchmark versions down to Elasticsearch 1.7! Here's an example:

esrally --pipeline=from-distribution --distribution-version=2.0.0

Note: Some older Python implementations may not be able to decompress 1.7.x Elasticsearch distributions. If that affects you, please try upgrading your Python installation.

New tracks

We have added a logging track. You can run it with esrally --track=logging. Note that you have to update Rally to version 0.3.0 as this track will only be supported by the latest version of Rally. Keep also in mind that this is quite a large data set: In total you need roughly 160 GB of free disk space to run this track. The new track is only tested and compatible with Elasticsearch 5 at the moment but support for older versions will be added in the next days.

Full changelog

If you have questions or feedback, please just post in the Elasticsearch forum.

Rally errors
(system) #2