Recently I worked on one customer issue that I would describe as “slave performance too slow”. During a quick analysis, I’ve found that the replication slave SQL thread cannot keep up while processing row-based events from the master’s binary log.
For example:
mysql> SHOW SLAVE STATUSG
*************************** 1. row ***************************
...
Master_Log_File: binlog.0000185
Read_Master_Log_Pos: 86698585
...
Relay_Master_Log_File: binlog.0000185
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
...
Exec_Master_Log_Pos: 380
Relay_Log_Space: 85699128
...
Master_UUID: …[Read more]