On 06/01/2011 12:18 PM, Luís Soares wrote:
> Hi Sven,
> Thanks for the review.
> Find some answers to your questions below.
> On 06/01/2011 11:14 AM, Sven Sandberg wrote:
>> Hi Luís,
>>> +--connection slave
>>> +call mtr.add_suppression ("Slave SQL: Table definition on master and slave
> does not match: Column 0 type mismatch.*");
>> Why suppress this? Don't we expect that nothing is written to the error log?
> There is still an error written to the error log:
> Slave SQL: Table definition on master and slave does not match: Column 0 type
> mismatch - received type 254, test.t1 has
> type 10, Error_code: 1535
> This is because ER_BINLOG_ROW_WRONG_TABLE_DEF is thrown at
> Now that you mention it, I wonder if we should add a parameter to
> compatible_with (..., bool suppress_error= false)
> What do you think ? Perhaps it is worth, since monitoring tools
> may be looking into the error log and things won't add up...
Adding a suppresion flag may not work (in case skip-all-errors is set
to a specific error code). Perhaps we can check ignored_error_code
before reporting the error. I.e., rli->report will set the error in the
diagnostics area and write to the error log. With this patch we take
care of the former, the latter is still there!