|Version 36 (modified by 13 years ago) ( diff ),|
Table of Contents
Replacing the simple SQLite database with MySQL allows greater scalability
- The databases directory contains what web2py knows about your database
- You need to set deployment_settings.base.migrate=True in
models/000_config.pybefore you update the code and switch back to deployment_settings.base.migrate=False later on.
- In case of database mismatches - check the databases directory for the appropriate .table file and change accordingly.
apt-get install mysql-server
optional, but highly recommended for easier GUI-based administration
apt-get install phpmyadmin
Install Python support for MySQL
apt-get install python-mysqldb
wget http://pypi.python.org/packages/source/M/MySQL-python/MySQL-python-1.2.3.tar.gz#md5=215eddb6d853f6f4be5b4afc4154292f tar zxvf MySQL-python-1.2.3.tar.gz cd MySQL-python-1.2.3 python setup.py install
For Windows try this binary:
Create the database
Database should be created within MySQL before Web2Py can see it to add the relevant tables:
mysql -u root -p CREATE DATABASE sahana; GRANT SELECT,INSERT,UPDATE,DELETE,CREATE,INDEX,ALTER,DROP ON sahana.* TO 'sahana'@'localhost' IDENTIFIED BY 'mypassword'; FLUSH PRIVILEGES;
Connection pools allow connections to be reused.
The number used is defined in
How many pools?
Whatever you choose, it always starts with one and grows with the number of concurrent requests up to the value of pools (the max number of concurrent pools). So pools should be the max number of concurrent requests you expect. That is the max value of:
<number of requests/second> / <average time per request in seconds>
For Production instances, give the MySQL database plenty of RAM:
innodb_buffer_pool_size = 2048M innodb_file_per_table innodb_thread_concurrency = 8
Index columns which are sorted on (especially on tables with many records) [we sort by name in most dropdowns, however most of the time this field is unique & so indexed automatically]:
CREATE INDEX name__idx on gis_location(name); CREATE INDEX first_name__idx on gis_location(first_name); CREATE INDEX middle_name__idx on gis_location(middle_name); CREATE INDEX last_name__idx on gis_location(last_name);
These should already be done automatically by
Possibly useful tips:
Can view the list of expensive queries by editing /etc/mysql/my.cnf and uncommenting/adding:
log_slow_queries = /var/log/mysql/mysql-slow.log long_query_time = 2
and restart mysql.
This will log queries that take longer than 2 seconds to run. View with:
tail -f /var/log/mysql/mysql-slow.log
By installing 'mytop', you can see real-time status on the database:
sudo apt-get install mytop mytop -uUSERNAME -pPASSWORD
type 'o' to force the longest-running queries to the top of the screen.
Outside of this, the usual tools for monitoring memory/cpu usage is effective.
Note that high mysql cpu usage may indicate a high io-wait condition, as many programs do not report iowait seperately from user / sys cpu usage. This hides the real issue. top is your friend. (iowait is time spent reading/writing to io; in this case, disk). High iowait can occur when attempting to view or sort massive quantities of data in eden. Using the above tips for increasing mysql ram usage will give mysql space to do sorts in memory.
In order to have a backup instance, configure replication from the Master node to a Backup node:
pico /etc/mysql/my.cnf #bind-address = 127.0.0.1 server-id = 1 log_bin = /var/log/mysql/mysql-bin.log binlog_do_db = sahana /etc/init.d/mysql restart mysql -u root -p GRANT REPLICATION SLAVE ON sahana.* to username@'sl.av.e.ip' IDENTIFIED BY 'mypassword'; FLUSH PRIVILEGES; SHOW MASTER STATUS;
Read log filename & position for configuring the slave (see below).
pico /etc/mysql/my.cnf server-id = 2 master-host = ma.st.er.ip master-user = username master-password = mypassword master-connect-retry = 60 replicate-do-db = sahana /etc/init.d/mysql restart mysql -u root -p SLAVE STOP; CHANGE MASTER TO MASTER_HOST='ma.st.er.ip', MASTER_USER='username', MASTER_PASSWORD='mypassword', MASTER_LOG_FILE='log_file', MASTER_LOG_POS=log_position; SLAVE START; crontab -e # m h dom mon dow command 55 * * * * /home/instance/scripts/automysqlbackup 55 * * * * /home/instance/scripts/sync-dir
Can configure the slave to adopt the same IP as the master using a Heartbeat:
If using Slicehost: