On Thu, 29 Apr 2010 08:40:19 -0700
Database System <database100@stripped> wrote:
> Is there anyone does replication on heavy duty database? There are a
> lot of updates on tables. When I replicate that server, I found the
> slave cannot continuely to connect to master, although Slave_I/O and
> SQL_I/O both are "Yes". It had so many times reconnect each second.
> Here is somehting from error log:
> 100429 9:03:50 [Note] Slave I/O thread: Failed reading log event,
> reconnecting to retry, log 'binlog.040932' position 26904210 100429
> 9:03:50 [Note] Slave: connected to master
> 'rep_name@host_name',replication resumed in log 'binlog.040932' at
> position 26904210 100429 9:03:50 [Note] Slave: received end packet
> from server, apparent master shutdown: 100429 9:03:50 [Note] Slave
> I/O thread: Failed reading log event, reconnecting to retry, log
> 'binlog.040932' position 26908777
> It is still replicating, but has failuer so many times per second,
> fills up error log. There is no network issue can be found. ping to
> master is normal.
Normal, means no packet loss over very long time, could the network be
some kind of saturated?
Is there any correlation between disconnects and database activity?
Does the master say something in the error log?