From: Rick James
Date: May 14 2012 9:00pm
Subject: RE: drop partitions
List-Archive: http://lists.mysql.com/mysql/227437
Message-Id: <2E7DD7ADE53B044C8C8BCD9C5829E1EB1485F37261@SP2-EX07VS01.ds.corp.yahoo.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Note that in many (too many) cases, _all_ partitions are opened, even if on=
ly one is really needed.

> -----Original Message-----
> From: louis liu [mailto:ylouis83@stripped]
> Sent: Monday, May 14, 2012 7:02 AM
> To: Johan De Meersman
> Cc: mysql@stripped; Rick James
> Subject: Re: drop partitions
>=20
> not actually ,  first partition to have been dropped   is no longer in
> use
> ,we dropped partition from old -> new
>=20
>=20
>=20
>=20
> 2012/5/14 Johan De Meersman <vegivamp@stripped>
>=20
> > ----- Original Message -----
> > > From: "Rick James" <rjames@stripped>
> > >
> > > If you have 14 partitions in each of 390 tables, and if you have
> > > most of the tables 'active', then you are possibly thrashing in the
> > > table_open_cache.
> >
> > A distinct possibility.
> >
> > > Compute (SHOW STATUS):
> > > Opened_tables / Uptime -- don't want more than a few per sec.
> > > Opened_files / Uptime -- ditto
> > > Opened_table_definitions / Uptime -- ditto
> >
> > Assuming you have anywhere near decent uptime, those are going to be
> > way too flattened to be of use. Trend monitoring is a necessity -
> look
> > at Munin or Cacti.
> >
> >
> > A random thought, though - could the first partition to have been
> > dropped not have been in active use, so the drop operation had to
> wait
> > for all other statements to end, and the other (older?) partitions no
> > longer in use, so could be dropped immediately?
> >
> >
> > --
> > Bier met grenadyn
> > Is als mosterd by den wyn
> > Sy die't drinkt, is eene kwezel
> > Hy die't drinkt, is ras een ezel
> >
>=20
>=20
>=20
> --
> Phone: +86 13918046970
> Email & Gtalk:  ylouis83@stripped
> Personal Blog: http://www.vmcd.org