Re: bzr commit into mysql-5.1 branch (mattias.jonsson:2732) Bug#38804

[Konstantin Osipov]

* Mattias Jonsson <Mattias.Jonsson@stripped> [08/09/08 17:41]:
>  2732 Mattias Jonsson	2008-09-08
>       Bug#38804: Query deadlock causes all tables to be inaccessible.
>       
>       Problem was a mutex added in bug n 27405 for solving a problem
>       with auto_increment in partitioned innodb tables.
>       (in ha_partition::write_row over partitions file->ha_write_row)
>       
>       Solution is to use the patch for bug#33479, which refines the
>       usage of mutexes for auto_increment.

I found nothing suspicious in the patch. Taking into account 
it's the fourth review, I am not requesting any minor changes.

Some considerations while looking at the patch (perhaps will be
useful in future):
 - if error inside a partition, it seems ha_partition should first 
 try to complete the operation on the rest of partitions, then
 return an error
 - an attempt to evaluate value of auto_increment_safe_stmt_log_lock 
 inside the storage engine is a violation of module borders. It
 seems this knowledge is specific to replication, so we should set
 this value somewhere inside decide_logging_format() function
 instead.

--