Skip to content

Latest commit

 

History

History
68 lines (52 loc) · 2.07 KB

README.md

File metadata and controls

68 lines (52 loc) · 2.07 KB

benchmarks

repository for benchmark scripts and such

Idea

This is (going) to be a set of scripts helping to build a test environment and test your Galera setups. We are using ansible to automate and distribute tasks. You don't need to log on any of the nodes as you are going to manage your cluster/testing from you local machine. So this repo is going to be your working directory for your ongoing tasks.

There are 4/5 modules planed

  • Provisioning (not implemented)
  • Installation (ongoing)
  • Testing (ongoing)
  • Visualising (not implemented)

Provisioning

This module is going to help you to launch instances in AWS or an OpenStack environment.

Installation

This module/scripts are going to install and configure you a

  • Galera cluster
  • Deploys the testing script on a separate machine

Now (20131226) you are going to have a cluster of $yournumber of galera nodes with $yournumber of testing/sysbench nodes pointing to one of the galera nodes.

We are going to support the Debian(Ubuntu) and RedHat OS family. Regarding Galera you can choose between Codership, MariaDB, Percona. ( Right now only Codership is supported on RehHat)

Testing

While you can install your testing hosts. We use sysbench 0.5 for the testing. You can invoke the script standalone or just run a (so called) play. Starting a play/test run you can configure your galera settings. A rolling restart is done for you automatically and result is stored in your /tmp directory

Format

The result is written in yaml and include

  • SHOW GLOBAL VARIABLES
  • SHOW GLOBAL STATUS (before and after each sysbench test)
  • sysbench results It is planned to put to put the environment (virtualisation and (ansible -m setup)) informations into that yaml to. (Maybe we switch to json also) The idea is to have as much informations about a run as we need even for later investigation.

Visualising

This is about extracting and visualise the informations. BIG TODO

WARNING

This is work in progress. It is going to be changed on a daily basis :)