MySQL Bugs: #28608: replication fail frequently

Bug #28608	replication fail frequently
Submitted:	23 May 2007 1:35	Modified:	27 May 2007 11:28
Reporter:	JinRong Ye	Email Updates:
Status:	Not a Bug	Impact on me:	None
Category:	MySQL Server: Replication	Severity:	S3 (Non-critical)
Version:	5.0.37	OS:	Linux (Red Hat Enterprise Linux AS release 4 (Nahant Update 2))
Assigned to:		CPU Architecture:	Any
Tags:	replication

Description:
I setup mysql replication between two linux server with the 5.0.37 version. The slave report error messages below frequently and got down:

070517 17:30:47 [Warning] "SELECT UNIX_TIMESTAMP()" failed on master, do not trust column Seconds_Behind_Master of SHOW SLAVE STATUS

or 

070523  7:49:52 [ERROR] While trying to obtain the list of slaves from the master '192.168.0.2:3306', user 'rep' got the following error: ''

or 

070522 21:23:32 [ERROR] Slave I/O thread: error reconnecting to master 'rep@192.168.0.2:3306': Error: 'MySQL server has gone away'  errno: 2006  retry-time: 2  retries: 100000000

There is no errors when i setup the replication on the same server with mysql 4.0.23 => mysql 5.0.37 and another replication under freebsd 6.2 server.

My replication configure in my.cnf:
server-id  = 21
master-host     = 192.168.0.2
master-user     = rep
master-password = rep
replicate-same-server-id    = FALSE
replicate-ignore-db  = mysql
replicate-ignore-db  = test
skip-slave-start
set-variable="slave-skip-errors=0,1007,1053,1062,1213,1158,1159"
log-slave-updates
master-connect-retry    = 2
master-retry-count      = 100000000
slave_net_timeout       = 1
log-bin     = /logs/mysql/logbin
relay-log   = /logs/mysql/relay
log-bin-index   = /logs/mysql/logbin.index
relay-log-index = /logs/mysql/relay.index
master-info-file    = /logs/mysql/master.info
relay-log-info-file = /logs/mysql/relay.info

How to repeat:
Unknown, sorry, but it is happening frequently.

Thank you for the report.

Please upgrade to last 5.0.41 version and examine if problem still exists.

Also 'MySQL server has gone away' error can show what something wrong with master. Please check master error log. Also please read change logs for MySQL since 4.0.23 untill 5.0.41, because there were several backward incompatible changes and your application can meet them.

there is no error log on master,thanks

Please turn on log-error setting in your master configuration file.

sorry,i mean there is no error message on master,i have turn on the log-error.

i set the global variables slave_net_timeout to 3(the original value is 1)，mysql report the error messages above become not so frequently

Thank you for the feedback.

Default value of slave_net-timeout is 3600. Also you can meet network problems. So I'll classify this report as "Not a Bug".