If your Slaves are not Masters, then Slaves do not need binary logging at all. You can put a cap on the amount of relay log space accumulated by a Slave. In order to throttle relay logs at 4G, add
relay_log_space_limit to /etc/my/.cnf on every Slave
and restart mysql
If you cannot set this, at least you should have some kind of alerting that does
SHOW SLAVE STATUS\G and check the value of
Relay_Log_Space (total bytes consumed by relay logs).
As for the Master, you could set
expire_logs_days to 1, but there is a severe warning I have for you...
If replication breaks, you have 1 day to fix it. Otherwise, a binary log on the Master may rotate away and you cannot run any CHANGE MASTER TO command to realign replication. I would leave
expire_logs_days at 3 on the Master.
If you have any overnight bulk processing to do, maybe should run the bulk processes on on the Master with
SET SQL_LOG_BIN=0; at the Start of the Session. This, of course, will not replicate to the Slave. You can perform the Same Bulk Load in Parallel to both Slaves.
Another thing you could do to manage the Master binary logs accumulation is this.
SHOW SLAVE STATUS\G on both Slaves. Look at
Relay_Master_Log_File. That represents the binary log on the Master whose last command was executed on the Slave.
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
1 row in set (0.00 sec)
In this example, Relay_Master_Log_File is mysql-bin.009590. All binary logs before this one can be removed from Master. You could run this on the Master:
PURGE BINARY LOGS TO 'mysql-bin.009590';
This will erase older logs and still leave replication in tact.
Binary Logs are files that serially compiles (like a FIFO queue) all completed SQL transactions as either a SQL statement or a row change. A relay log is a file that collects binary log entries from a remote server (aka Master).
In MySQL Replication
- Master must have its binary logs enabled
- Slave compiles relay logs
- When all SQL in a relay log is processed, it is deleted
- On a Slave, when there is more that one relay log on a DB Server, it may indicate replication is falling behind because the IO thread is collecting SQL from a Master faster that the SQL thread can process the relay logs.
- Using relay_log_space_limit prevents replication from piling up and potentially filling up a disk. Relay logs rotate out based on rule #3
- It is possible for a DB Server to be both a Master and Slave. That's the only circumstance under which a Slave must have binary logs enabled. In that scenario, a DB Server will have both Binary Logs and Relay Logs.
If you failover to a Slave, and you want to Make it a Master
- service mysql stop
log-bin=mysql-bin to /etc/my.cnf on the Slave
- service mysql start
You will have to setup replication of other Slaves to the newly promoted Master and make sure the data on the Slave match up with the newly promoted Master
UPDATE 2012-08-13 17:47 EDT
According to the MySQL Documentation on
relay-log option, you should define it. Here is why:
Due to the manner in which MySQL parses server options, if you specify this option, you must supply a value; the default basename is used only if the option is not actually specified. If you use the --relay-log option without specifying a value, unexpected behavior is likely to result; this behavior depends on the other options used, the order in which they are specified, and whether they are specified on the command line or in an option file. For more information about how MySQL handles server options, see Section 4.2.3, “Specifying Program Options”.