[DOCS] Shadowdb and Table locking issues

Max Kanat-Alexander mkanat at bugzilla.org
Sun Feb 3 02:54:06 UTC 2008

On Sat, 2 Feb 2008 17:04:26 -0500 "Sam Folk-Williams"
<sam at predisposition.com> wrote:
> Hi Max - OK so it looks like the switch was made to InnoDB to avoid
> this in 3.2, correct? But 3.0 and below still uses MyISAM (and hence
> locking).

	That's right.

> So should this whole shadow DB thing be removed for 3.2? Or
> is there still some benefit to using it?

	I really have no idea at this point, actually.

> OK but for 3.2 this is no longer correct: "These types are slower than
> the standard
> type, and Bugzilla does not yet take advantage of features such as
> transactions which would justify this speed decrease". It sounds like
> now Bugzilla /does/ use these features and InnoDB, yeah?

	Yes, kind of. Since we aren't MySQL-specific, we don't use the
MySQL-specific row-locking SQL, or at the very least we haven't yet
identified any race conditions that would make that necessary. There's
a certain sort of row-level locking that happens automatically inside
our transactions.

	Anyhow, that was me being all technical, but the simple answer
is "Yes, that section is no longer needed."

Competent, Friendly Bugzilla and Perl Services. Everything Else, too.

More information about the developers mailing list