MySQL Event Processor

Daemon interface for handling MySQL binary log events. You may implement your own handlers to handle events like INSERT, UPDATE and DELETE with data of the affected rows.

Requirements

• MySQL >= 5.5
• Python2 >= 2.7
• python-mysql-replication: Pure Python Implementation of MySQL replication protocol build on top of PyMYSQL.
• see mysqlevp/handler/requirements.txt for requirements of built-in handlers

Installation

This package is currently in alpha version and not registered in PyPI. You can only install it with github.

pip install "git+https://github.com/jffifa/python-mysql-eventprocessor.git@master"

MySQL Master Server Settings

MySQL Event Processor daemon behaves like a MySQL replication slave server, so you need to configure MySQL Master Server properly for binary log settings.

Here is an sample configuration for my.cnf

[mysqld]
server-id        = 1
log_bin          = /var/log/mysql/mysql-bin.log
expire_logs_days = 10
max_binlog_size  = 100M
# the binlog-format settings strongly affects implementations of
# the replication server
# be sure to set to "row" to receive INSERT, UPDATE and DELETE events.
# http://dev.mysql.com/doc/refman/en/binary-log.html
# http://dev.mysql.com/doc/refman/en/replication-formats.html
binlog-format    = row

MySQL Replication User

The daemon will connect to MySQL master server as a replication slave, so you must create slave user to connect master server.

Otherwise, you may use "root" user, which is not recommended in production environment.

Usage

Run as a Normal Program

1. Choose an event handler or implement your own handler. See Handler Developing for instructions in developing your own handlers.

2. Write mysql connection settings.

   mysql_conn_settings = {
       'host':'127.0.0.1',
       'port':3306,
       'user':'root',
       'passwd':'password',
   }
   

   Be sure the account have appropriate privilege.

3. Assign a unique replication server ID to the daemon. Do not conflict with other slave connected to the master server.

4. Choose a path the daemon dump its status info to. Be sure that the daemon has write and read access to the directory of the path.

5. Import mysqlevp.deamon.MysqlevpWrapper and create an instance with the params below. And then just run.

   from mysqlevp.daemon import MysqlevpWrapper
   mysqlevp = MysqlevpWrapper(
       mysql_conn_settings,
       replication_server_id=196883,
       my_event_handler,
       '/tmp/mysqlevp.dmp')
   mysqlevp.run()
   

See examples/write_to_console.py for details and other options.

Run as a Daemon Process

Use process manager like supervisor to control it, or write your own python DaemonContext.

Handler Developing

You should write a class derived from mysqlevp.handler.base.IEventHandler, and implement the following 3 functions:

from mysqlevp.handler.base import IEventHandler

class MyHandler(IEventHandler):
    def on_insert(self, ev_id, ev_timestamp, schema, table, affected_rows):
        ...

    def on_update(self, ev_id, ev_timestamp, schema, table, affected_rows):
        ...

    def on_delete(self, ev_id, ev_timestamp, schema, table, affected_rows):
        ...

Read mysqlevp/handler/base.py for param documents and the built-in handler mysqlevp/handler/console.py for example.

TODO

• Python3 support: Though using package six for compatibility, this project has not been tested under Python3 yet.
• better logging utils
• MySQL Global Transaction Identifier support.
• Allow wild chars in table filters for database or table partions. (CREATE TABLE t2 LIKE t etc.)

Similar Works

• meepo: Meepo is event sourcing and event broadcasting for databases.