MySQL Bugs: #64562: replicate-do-db: inconsistency between actual behaviour and docs

Bug #64562	replicate-do-db: inconsistency between actual behaviour and docs
Submitted:	6 Mar 2012 10:32	Modified:	23 Mar 2012 12:11
Reporter:	Denis Kukharev	Email Updates:
Status:	Duplicate	Impact on me:	None
Category:	MySQL Server: Documentation	Severity:	S3 (Non-critical)
Version:	5.5	OS:	Any
Assigned to:	Jon Stephens	CPU Architecture:	Any
Tags:	inconsistency, replicate-do-db, replication, table options

Description:
Hi!
A common way to set up simple replication at slave side is to provide several replicate-do-db options. Let's assume in addition that we have no table options at all. All works fine and reasonable, namely if an event's DB doesn't match any one provided in replicate-do-db list, the event will be rejected.

However, this doc page http://dev.mysql.com/doc/refman/5.5/en/replication-rules-table-options.html says, first:
"The slave checks for and evaluates table options only if no matching database options were found" (that is the case) and  second, that if no table option is given (that is the case, too) then the update WILL be executed. This is applicable for the activity diagram provided on the page and its step-by-step description below.

I believe (and hope) the bug is in the documentation itself rather than in MySQL functionality.

How to repeat:
1. set up simple master-slave replication with only DB-level (replicate-do-db) options on slave.
2. execute SBR statement on master having default DB set to not match any DB given in slave's replicate-do-db list.
3. compare slave behaviour to the documentation

Suggested fix:
Update the documentation

Thank you for the report.

Verified as described: if look at the diagram only one can think if there is no replicate-*-table option is given statement will be executed independently if replicate-do-db option was set or not.

Sveta, having taken the text above the diagram into account (about pre-checking db-level options) as I've described and everyone should :), one can think an update is to be executed, too.

Duplicate of BUG#60188. 

(NB-That bug refers to the binlog versions of these options, but we will review the docs for all the the *db* filtering options in the course of fixing that issue.)