Multiple table spaces eh? Funky! Of course, as you've no doubt become
aware, I am one of these annoying
people that has many questions to ask the list the moment something like
this comes along! Please feel
free to ignore these questions though, as we're all hanging out for 4.1.1.
1. Are there any performance implications (either way) with the use of
multiple table spaces? Are there any
other implications that you think are worthy of note (other than the
obviously easier backup options).
2. Is there any way that clustering can still happen with multiple table
Heikki Tuuri wrote:
>InnoDB is a MySQL table type which provides transactions, row-level locking,
>non-locking consistent SELECT (multiversioned concurrency control), foreign
>key constraints, and a non-free hot backup tool for backing up InnoDB
>InnoDB is included in all MySQL-4.0 and 4.1 downloads, and also in the MySQL
>Pro commercial, non-GPL MySQL license.
>Release 4.0.16 is a bugfix release of the stable 4.0 branch. There are a few
>known outstanding bugs in InnoDB-4.0.16, but their fixing has been delayed,
>because we are allocating all free resources to preparing the upcoming
>A sneak peek of MySQL-4.1.1:
>NOTE: if you upgrade to InnoDB-4.1.1, you cannot downgrade any more! That is
>because earlier versions of InnoDB are not aware of multiple tablespaces.
>The public BitKeeper source tree of 4.1 now supports multiple tablespaces
>for InnoDB. You can enable them with the line
>in the [mysqld] section of my.cnf. Then InnoDB stores each table into its
>in the database directory where the table belongs. This is like MyISAM does,
>but MyISAM divides the table to a data file tablename.MYD and the index file
>tablename.MYI. For InnoDB, both the data and the indexes are in the .ibd
>If you remove the line, then InnoDB creates tables in the ibdata files
>InnoDB always needs the 'system tablespace', .ibd files are not enough. The
>system tablespace consists of the familiar ibdata files. InnoDB puts there
>its internal data dictionary and undo logs.
>You CANNOT FREELY MOVE .ibd files around, like you can MyISAM tables. This
>is because the table definition is stored in the InnoDB system tablespace,
>and also because InnoDB must preserve the consistency of transaction id's
>and log sequence numbers.
>You can move an .ibd file and the associated table from a database to
>another (within the same MySQL/InnoDB installation) with the familiar RENAME
>RENAME TABLE olddatabasename.tablename TO newdatabasename.tablename;
>If you have a 'clean' backup of an .ibd file taken from the SAME
>MySQL/InnoDB installation, you can restore it to an InnoDB database with the
>ALTER TABLE tablename DISCARD TABLESPACE; /* CAUTION: deletes the current
>.ibd file! */
><put the backup .ibd file to the proper place>
>ALTER TABLE tablename IMPORT TABLESPACE;
>'Clean' in this context means:
>1) There are no uncommitted modifications by transactions in the .ibd file.
>2) There are no unmerged insert buffer entries to the .ibd file.
>3) Purge has removed all delete-marked index records from the .ibd file.
>4) mysqld has flushed all modified pages of the .ibd file from the buffer
>pool to the file.
>You can make such a clean backup .ibd file with the following method.
>1) Stop all activity from the mysqld server and commit all transactions.
>2) Wait that SHOW INNODB STATUS\G shows that there are no active
>transactions in the database, and the 'main thread' of InnoDB is 'Waiting
>for server activity'. Then you can take a copy of the .ibd file.
>Another (non-free) method to make such a clean .ibd file is to
>1) Use InnoDB Hot Backup to backup the InnoDB installation.
>2) Start a second mysqld server on the backup and let it clean up the .ibd
>It is in the TODO to allow moving clean .ibd files also to another
>MySQL/InnoDB installation. That requires resetting of trx id's and log
>sequence numbers in the .ibd file.
>InnoDB changelog for 4.0.16:
>* Fixed a bug: contrary to what was said in the manual, in a locking read
>InnoDB set two record locks if a unique exact match search condition was
>used on a multi-column unique key. For a single column unique key it worked
>* Fixed a bug: if one used the rename trick #sql... -> rsql... described in
>section 15.1 of http://www.innodb.com/ibman.html to recover a temporary
>table, InnoDB asserted in row_mysql_lock_data_dictionary().
>* There are several outstanding non-critical bugs reported in the MySQL bugs
>database. Their fixing has been delayed, because resources are allocated to
>the upcoming 4.1.1 release.