List:Maria Storage Engine« Previous MessageNext Message »
From:Michael Widenius Date:October 31 2008 9:56pm
Subject:Re: bzr commit into MySQL/Maria:mysql-maria branch (monty:2672)
View as plain text  

Sanjay, what is the best list to discuss changes in the storage engine

I think this should be done one some public list, but internals may be
general for this.

Anyway, Guilhem recently pointed out a small change that I did that
you should know about:

>>>>> "Guilhem" == Guilhem Bichot <Guilhem.Bichot@stripped> writes:

Guilhem> Hello Monty,
Guilhem> Michael Widenius a écrit :
>> #At bzr+ssh://
>> 2672 Michael Widenius	2008-10-04
>> Fix for bug#39226 Maria: crash with FLUSH TABLES WITH READ LOCK after LOCK
>> The problem was that we didn't inform the handler that we are going to close
> tables
>> that are locked and may have (at least in mearia) be part of an active
> transaction

>> per-file messages:
>> include/my_base.h
>> Added HA_EXTRA_PREPARE_FOR_FORCED_CLOSE for signaling the handler that the file
> will be forced closed

In other words, I added to the MySQL-maria tree (which is now merged
to 6.0) a new extra flag: HA_EXTRA_PREPARE_FOR_FORCED_CLOSE.

This is called when MySQL is going to force close of table that is
part of a transaction or locked.

Maria needs to know this because it tracks all tables that are part of
a transaction and it will normally not allow one to close a table
that is still in use. By calling extra() Maria will be signaled that
it should remove the table from being a part of the current transaction

This change is safe as an engine is allowed to ignore any extra
call().  This call only gives the engine more information what's going
on at the server level.

bzr commit into MySQL/Maria:mysql-maria branch (monty:2672) Bug#39226Michael Widenius4 Oct
  • Re: bzr commit into MySQL/Maria:mysql-maria branch (monty:2672)Bug#39226Guilhem Bichot6 Oct
    • Re: bzr commit into MySQL/Maria:mysql-maria branch (monty:2672)Bug#39226Michael Widenius31 Oct