From: Wei He
Date: November 1 2000 2:18am
Subject: Re: A suggestion on Replication
List-Archive: http://lists.mysql.com/internals/91
Message-Id: <Pine.LNX.4.21.0011010957200.24490-100000@mail.ied.ac.cn>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Thu, 26 Oct 2000, Christian Rabe wrote:

Thanks for your instruction. I didn't have and change to try again until
todayb I found my replication being interrupted again. In my
environment, your first and third cases can be excluded. 

Instead of using temporary tables, I use variables like @ID, @tempid...
Since the connection between my master and slave is poor, I wonder once 
the connection is broken, the @ID will lose its value when reconnected.
I this case, hacking of slave.cc will not only of no help but dangerous.
Am I right? 

Wei He





> Hi,
> 
> To 1: To accomplish this, you have to disable the lines 926-933 in the
> sql/slave.cc (3.23.26)
> To 2: This is already the case. Only successful master-querys are logged.
> There are 3 cases, so far I know, when the
>         slave gets a query to break on.
>         First: there are Tables on the master that are not on the slave and
> you write to them.
>         Second: You use temporary tables,which are not temporary on the
> slave and therefore not deletet afterward.
>         So the Slave breaks the moment you try to recreate it on the master.
>         In this two cases the command "SET SQL_LOG_BIN = 0" can help you,
> which switches the logging off.
> 
>         Third: You make a flush slave without a former flush master. The
> slave reads
>         then all existing logfiles anew which can lead to problems with
> duplicate entries on the slave and BANG!
> 
> I too had to use the dirty way of disabling the error with the drawback that
> when there really is an error you won't notice it.
> 
> Regards,
> 
> Christian "Maverick" Rabe
> IT-Development
> wallstreet:online AG
> ------------------------------------------
> http://www.wallstreet-online.de
> Tel.:  03 37 01 5 29 16
> Fax.:  03 37 01 5 29 19
> 
> 
> ----- Original Message -----
> From: Wei He
> To: internals@stripped
> Sent: Thursday, October 26, 2000 4:45 AM
> Subject: A suggestion on Replication
> 
> 
> Hi,
> 
> According to my experience, the replication will be broken when the slave
> executes a error sql command. Can this be changed to:
> 
> 1. The slave ignores the error and continues the replication;
> 2. The master only bin-logs sql commands that succeed.
> 
> 
> Wei He
> 
> 
> ---------------------------------------------------------------------
> Please check "http://www.mysql.com/Manual_chapter/manual_toc.html" before
> posting. To request this thread, e-mail internals-thread55@stripped
> 
> To unsubscribe, send a message to the address shown in the
> List-Unsubscribe header of this message. If you cannot see it,
> e-mail internals-unsubscribe@stripped instead.
> 
> 
> ---------------------------------------------------------------------
> Please check "http://www.mysql.com/Manual_chapter/manual_toc.html" before
> posting. To request this thread, e-mail internals-thread56@stripped
> 
> To unsubscribe, send a message to the address shown in the
> List-Unsubscribe header of this message. If you cannot see it,
> e-mail internals-unsubscribe@stripped instead.
> 
>