Query Improvement (was Re: Bugzilla Contribution Process (was RE: New language discussion?))

[Max Kanat-Alexander]

On Fri, 2 Nov 2007 16:06:22 -0700 David Marshall
<dmarshal at yahoo-inc.com> wrote:
> Furthermore, the main problem that we are solving this way manifests  
> itself really only when replicating to a shadow database.  If a very  
> slow query is processing when a replication statement shows up, the  
> replication must wait for the slow query.  Then, any new queries
> must wait for the replication to finish.  This can lead to a horrible
> logjam.

	This can probably be solved in 3.2 code by using transactions
and allowing searches to do READ UNCOMMITTED transactions. Locks will
be eliminated in 3.2 (I hope!) so there shouldn't be any more blocking.

	-Max
-- 
http://www.everythingsolved.com/
Competent, Friendly Bugzilla and Perl Services. Everything Else, too.