MySQL Bugs: #2917: binlog-*-table needed

Bug #2917	binlog-*-table needed
Submitted:	21 Feb 2004 11:04	Modified:	12 Sep 2022 14:44
Reporter:	Przemyslaw Popielarski	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Server: Replication	Severity:	S4 (Feature request)
Version:	4.0.18	OS:	Linux (Linux 2.4.x x86)
Assigned to:	Assigned Account	CPU Architecture:	Any

Description:
Well, binlog-*-table would be very appreciated. I'm trying to minimize traffic between master and slave, and now as of 4.0.18 even with replicate-ignore-table the slave gets the binary log from master and then parses it and does nothing more. It's just useless network traffic.
regards, premax@hot.pl

How to repeat:
put in master's /etc/my.cnf:
 binlog-ignore-table=DB.TABLE
and unfortunately is does not work :-)

Suggested fix:
Add options:
 --binlog-ignore-table=DB.TABLE
 --binlog-do-table=DB.TABLE
 --binlog-wild-ignore-table=DB.TABLE
 --binlog-wile-do-table=DB.TABLE
to be supported on master.

I can confirm.
For example, we smashed around 500 GB of replication data in 8 hours. The slaves just skip the tables in "replicate wild ignore table" but like you said, useless network traffic.

And if you have a lot of data moving around on the master which doesn't need replication it's producing high IOwait load and network load (both on master and slave).

has this been implemented in later versions.

Has this been added to any later versions of MySQL?

I have 5.0.27-community, and I don't see this option in the documentation.

Yes! please i need that feature too.
I have 4 servers, 1 master and 3 slaves. They are not in a LAN, they are in different cities. I configure the my.ini in the slave for replicating just a few tables, not all the database and very often i have replication crashes because every modification on Master is sent to the slaves, not only the tables i want. The more common error is because sql statements (until now, they´re all for tables other than i want to replicate) are not crc verified and arrive to slaves with syntax errors (wrong data). 
I think that is necesary some binlog-*-table option, for 2 reasons (may be there are more):
1) the type of crash i explained on top
2) because without that option and with big servers (that move a lot of data daily) there are too much (senseless) network traffic.

Please sorry my english, i hope you can understand
I'm working with 5.0.45-community-nt-log, MySQL Community Edition (GPL)

This feature would be great to have

count me in for the feature!!!

So much in for the feature!

Just wanted to add. I have a situation where LOAD DATA INFILE loads 16Mb files into tables. I only need to do this operation once on master and never replicate, so really cringing on the idea that I'll have to drag all that data load over the network. 

I think I might have found a work around with binlog-ignore-db though, but really this feature could be extremely handy :)

Still not there after 6 years? =(

Yeah, still waiting :>

Anybody know if this was ever implemented? I am in the same boat. 

I have a bunch of nightly work performed on "load" tables on the MASTER. Don't want/need this replicated across. Even with 'SQL Thread' turned off, there is a significant IO/load/performance impact on the SLAVE (during these operations).

Have NOT tried adding "replicate-wild-ignore-table=myDB.myLoadTable" to SLAVE. I believe this won't matter since I have the 'SQL Thread' off anyways during these operations....can anyone confirm this?

Just wanted to add that I ended up putting in "SET SQL_LOG_BIN = 0" before all load operations...this appears to have addressed the performance impacts on the slave

Here's my vote for binlog-wild-ignore-table......replicate-wild-ignore-table is not enough for 2 different clients of ours, as their binlogs are huge (bigger than their data!  One client has a 33G database, and every day produces 60 Gb of binlogs!).

Is this a particularly difficult feature?  It seems like many folks want it.

I would also like to see this feature

A particular use case I have in mind for something like this is a moodle app that writes a heavy amount of (session) data to a transient table.  I'd like to be able to make it an in memory table and not care about any sort of binary logging or replication.  As it stands I have to have 10x as much space for the binary logging for that data as I do for the actual data since I do want to replicate everything else.

Thanks,
Brian

+1.  This was requested over 10 years ago.  Please add in this feature.  It would be very useful.

I second that motion.
This feature will greatly reduce lag in my replication setup.

+1 
This feature is very important whenever you use the binlogs as the CDC source for ETLs, as sometimes we just want to ignore a table altogether.