Slave I/O thread dies very often

Question

I have MySQL Replication working between two servers in master-master mode.

Replication has been working from day 1.

The IO Thread and the SQL Thread has never died. However, the Slave SQL thread dies very, very often.

The log only says:

120913 17:58:50 [Note] Slave I/O thread killed while reading event
120913 17:58:50 [Note] Slave I/O thread exiting, read up to log 'mysql-bin.004109',
    position 101146947
120913 17:58:50 [Note] Error reading relay log event: slave SQL thread was killed

I've Googled for the problem, but found no indication of what to look for or how to resolve the issue.

What could be causing the thread to die?

Answer 1

When the IO Thread dies, it usually dies for these reasons

• STOP SLAVE;
• STOP SLAVE IO_THREAD;
• Network Issues

The first two reasons are obvious, but many have been victimized by network connectivity. For example, if there are enough dropped packets along the traceroute of the IP you are using as the Master, the IO Thread will just die off. By default, mysql attempts to reconnect to the Master every 60 seconds.

The mysqld process does not record network level issues, but a dying IO Thread is usually a dead giveaway (no pun intended). If this is happening often, go to work troubleshooting the traceroute on that IP (perhaps ifconfig on that Master IP).