On 08/05/17 11:48, Phil Stracchino wrote:
> Kern,
> Regarding the batching for attribute data - based on your explanation of
> how it works, I'm at a loss to understand why the person who wrote the
> MySQL code thought that was a good idea in the first place.  It seems
> the only thing it accomplishes is to write all of the attribute data
> TWICE.  I need to look at the code more, but at this point I'm really
> thinking that the entire batching idea was ill-conceived from the start.
>  It's possible it MIGHT have once been a good idea IF using legacy
> MyISAM tables, but honestly, the world would be better off had the
> MyISAM storage engine died with MySQL 4.1 in the late 90s.

....And this is where I facepalm.  Because I should have thought of this

You see, there's this configuration variable, --enable-batch-insert,
that is on by default.  And because it's on by default, I forgot it was
even there at all.

I'm testing right now and the Director works just fine with a Galera
cluster if it's compiled with --DISABLE-batch-insert.

In fact, not only does it WORK that way, it works *BETTER*, whether
running against a Galera cluster *OR* against a standalone mysqld.

I'm going to make the possibly radical suggestion that if using
--with-mysql, batch-insert should be DISABLED by default.  Because after
another dig through the code, I am now convinced that with a modern
MySQL version configured even approximately competently, the
batch-file-insert feature is not only of no benefit whatsoever, it is
actively *harmful* to both performance and functionality, and should not
be used.

  Phil Stracchino
  Babylon Communications
  Landline: +1.603.293.8485
  Mobile:   +1.603.998.6958

Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Bacula-devel mailing list

Reply via email to