Hello,
It's the continuation of the post
http://forums.mysql.com/read.php?26,377336 (sorry I can't reply on the previous post for an unknow reason from me ...)
My problem has occured again: the slave on the host #2 has crashed just after the flush logs.
I use a dual replication between 2 hosts : each host is master and slave for the other.
The version is "5.0.67" on both hosts. Tables are all MyISAM.
The flush is done with the command: FLUSH USER_RESOURCES
Same error as previously has been recorded in the errors log:
100812 0:30:01 [ERROR] next log error: -1 offset: 51 log: /u3/local/mysql/slave/repl/host2-relay-bin.000064 included: 1
100812 0:30:01 [ERROR] Error reading relay log event: Error purging processed logs
When the slave has been stopped, it was pointing exactly at the end of the relay bin log.
SLAVE STATUS displays:
Relay_Log_File: host2-relay-bin.000064
Relay_Log_Pos: 7058484
%mysqlbinlog host2-relay-bin.000064 | tail
/*!*/;
# at 7058434
#100812 0:30:01 server id 6 end_log_pos 7058484 Rotate to host2-relay-bin.000065 pos: 4
DELIMITER ;
# End of log file
ROLLBACK /* added by mysqlbinlog */;
/*!50003 SET COMPLETION_TYPE=@OLD_COMPLETION_TYPE*/;
The next relay bin log "host2-relay-bin.000065" has been created normally by the IO Thread but it has been never read by the SQL Thread.
One more thing weird : in the MySQL errors log I have seen that the parameter values of "CHANGE MASTER" are repeated many times (about every 5 secondes).
This problem occurs in average 3 or 4 times by month.
Each time I must reset and restart the slave (and sometimes restart the mysqld server like this morning).
I get 2 kind of error, the error above and the following:
100818 0:35:01 [ERROR] I/O error reading the header from the binary log, errno=-1, io cache code=0
Thank you for your help,
~~~
Eric