Hysen Labs
Open-source project
HustAIsGroup/CDBTune avatar
HustAIsGroup

CDBTune

CDBTune

100 stars40 forksPythonLicense varies
01
DEEP OPEN-SOURCE ANALYSIS

CDBTune: a two server setup guide for a tuning model

CDBTune's README walks through a two server environment for training a database tuning model with AutoTuner: one machine running MySQL 5.6 under sysbench load, the other running the training, in Chinese.

02
DEEP OPEN-SOURCE ANALYSIS

The environment runs on two servers

The README is written in Chinese and reads like an internal deployment note. It specifies a two server setup: CDBTune1 at one IP runs sysbench 1.0 and handles the network training plus active stress testing, while CDBTune2 runs MySQL 5.6 as the database server. Both machines use Ubuntu 16.04 with Python 2.7 and the same example user and password. The environment table is exact, down to IP addresses and credentials.

03
DEEP OPEN-SOURCE ANALYSIS

The setup is a step by step list

Setup follows a numbered list. The AutoTuner training project goes into the user's home directory on both servers, with a note to keep the project name unchanged. MySQL 5.6 installs on the database server, with the version called out repeatedly, a root password set, and the service started. Sysbench initialises the test database, creating eight tables of one million rows each. Dependencies install through pip, explicitly not pip3, on both machines. Then come the edits: a start train script modelled on the server script, MySQL connection details in the config file, username paths corrected in the MySQL module, and the MySQL config replaced with defaults drawn from the knobs file. After training, an evaluate script runs with the model parameter prefix passed in.

04
DEEP OPEN-SOURCE ANALYSIS

A troubleshooting section collects the pain points

A question and answer section gathers the errors people hit. MySQL failing to connect through its socket, a prepare script complaining about an unexpected operator, and MySQL-python failing to build all get answers, with one fix involving system packages and easy install before trimming the requirement file. A gcc error is blamed on missing dependencies. The README treats these as known speed bumps rather than mysteries.

05
DEEP OPEN-SOURCE ANALYSIS

Editorial conclusion

CDBTune's README is a Chinese deployment guide for a two server database tuning environment: sysbench on one box, MySQL 5.6 on the other, and AutoTuner trained against them. A troubleshooting section covers the common install errors.

06
DEEP OPEN-SOURCE ANALYSIS

Official sources

07
Community notes

Community notes