From: Sabika Makhdoom Date: February 19 2013 11:13pm Subject: Re: Upgrading form mysql 5.0.90 to 5.5 or 5.6 List-Archive: http://lists.mysql.com/mysql/229014 Message-Id: <4F9A05DE-4988-4C0E-ACC9-F74BB07EB6C4@gmail.com> MIME-Version: 1.0 (1.0) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: quoted-printable Use replication as your fail over and why not percona's xtrabackup or lvm ty= pe backup if you need a backup? Sabika=20 On Feb 19, 2013, at 1:20 PM, Reindl Harald wrote: > surely >=20 > * use "mysql_upgrade -u root -p" after EACH update > * upgrade regulary >=20 > we went from MySQL 3.x to 5.5.30 until know without > any dump and here are around 5000 tables >=20 > Am 19.02.2013 22:12, schrieb Divesh Kamra: >> Is there any better way for grade MySQL version without taking backup wit= h mysqldump >>=20 >> Or if there any tool for this=20 >>=20 >> R's >> DK >>=20 >> On 16-Feb-2013, at 16:07, Reindl Harald wrote: >>>=20 >>> Am 16.02.2013 09:42, schrieb Manuel Arostegui: >>>> 2013/2/15 Reindl Harald > >>>>=20 >>>> "our database is 400 GB, mysqldump is 600MB" was not a typo and you >>>> honestly believed that you can import this dump to somewhat? >>>>=20 >>>> WTF - as admin you should be able to see if the things in front >>>> of you are theoretically possible before your start any action >>>> and 1:400 is impossible, specially because mysql-dumps are >>>> ALWAYS WAY LARGER then the databasses because they contain >>>> sql-statements and not only data >>>>=20 >>>> That's not completely true. If you have a poor maintained database or j= ust tables with lot of writes and deletes >>>> and you don't periodically optimize it - you can end up with lot of bla= nk spaces in your tables which will use _a >>>> lot_ of space. If you do a "du" or whatever to measure your database si= ze...you can get really confused. >>>> mysqldump obviously doesn't backup blank spaces and once you get rid of= them, your database will use much less space. >>>=20 >>> ok, normally i expect there is a admin and doing his job >>> especially for large datasets >=20