-
Notifications
You must be signed in to change notification settings - Fork 81
Description
I have a WordPress setup with 5 dedicated database servers. 1 master for writing only and 4 slaves for users to read from. All servers have the same hardware configuration (Xeon 20/40 cores / 192 GB RAM) and my.cnf except for the server identification.
I have installed HyperDB and now I am trying Ludicrousdb and I find the same drawback: Load balancing is not real.
The first server always supports most of the work and the rest have a much lower load, causing the infrastructure to not be used.
What solution can there be?
!includedir /etc/my.cnf.d
[mysqld]
sql_mode=ERROR_FOR_DIVISION_BY_ZERO,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION
bind-address = 0.0.0.0
relay_log = /var/log/mariadb/mysql-relay-bin
relay_log_index = /var/log/mariadb/mysql-relay-bin.index
server-id = 3
replicate-do-db=loquosex2013
log_bin = /var/log/mariadb/mysql-bin.log
max_binlog_size = 10G
binlog_cache_size = 16M
replicate-ignore-db = apsc
replicate-ignore-db = mysql
replicate-ignore-db = information_schema
replicate-ignore-db = performance_schema
replicate-ignore-db = phpmyadmin
replicate-ignore-db = psa
replicate-ignore-db = roundcubemail
local-infile=0
max_connections = 3000
wait_timeout = 20
interactive_timeout = 100
skip-name-resolve=1
table_definition_cache=700
table_open_cache = 3000
tmp_table_size = 18M
max_heap_table_size=18M
join_buffer_size=1M
key_buffer_size = 2M
innodb_buffer_pool_size= 6G
innodb_log_file_size = 1G
innodb_log_buffer_size = 20M
thread_cache_size = 320
aria_pagecache_buffer_size = 256M
performance_schema = ON
