#Omnigraffle subscript manual#
#Omnigraffle subscript full#
#Omnigraffle subscript free#

The FIL header (and footer) is common to all page types and contains primarily information about the page itself.Īdditional information follows depending on the page type for INDEX pages the following information is dumped: The initial line will tell you which class is handling this page: $ innodb_space -f test/t.ibd -p 3 page-dump The innodb_ruby library initially parses pages using a minimal Innodb::Page class, and then using the type field present in the common header optionally hands off the different page types to specialized classes (such as Innodb::Page::Index for type INDEX) for further parsing.Ī good page to start looking would be the first INDEX page, which is the root node of the index tree for the test table created above, and is located at page 3: It currently leans heavily on the typical Ruby pretty-printer module pp to print the structures - that would be a great thing to clean up in the future. The page-dump mode dumps everything it knows about a single page.

#Omnigraffle subscript full#

(Keep in mind this has no relation to how full the index pages are, that is another matter.) Examining a single page

#Omnigraffle subscript free#

Pages may be allocated to a file segment but currently unused (type FREE (ALLOCATED)), so “fill_factor” will show the ratio of used to unused. $ innodb_space -f test/t.ibd space-indexesĮvery index has an “internal” file segment, used for non-leaf pages, and a “leaf” file segment, used for leaf pages. Without getting into too many of the InnoDB internal implementation details, you see some of InnoDB’s bookkeeping structures ( FSP_HDR, IBUF_BITMAP, and INODE pages), actual table data ( INDEX pages), and free space ( FREE (ALLOCATED) pages).Ī listing of space consumed, in pages, by each index (actually each “file segment”, or FSEG for each index) can be fairly interesting as well: $ innodb_space -f test/t.ibd space-page-type-regions One of the most high-level overviews possible with innodb_space is space-page-type-regions, which prints one line per contiguous block of a given page type: (Note that if you’re trying this at home, you’ll want to watch SHOW GLOBAL STATUS LIKE 'Innodb_buffer_pool_pages_dirty' to wait for all dirty pages to be flushed before proceeding, since the tools below will be accessing the tablespace file on disk, with no coordination with a running InnoDB instance.) Examining a tablespace file This should generate a table of 1 million rows (inserted in random order to make things more interesting), of about 48MiB, or 3,071 16KiB pages. M.query('INSERT INTO t (i) VALUES (# rows.' if index % 10000 = 0 M.query('CREATE TABLE t (i INT UNSIGNED NOT NULL, PRIMARY KEY(i)) ENGINE=InnoDB') Create and populate a very simple table with a small bit of Ruby: Make sure you’re running a new enough server (MySQL 5.5 is good) with Barrracuda tables and that you have innodb_file_per_table enabled.

Usage: innodb_space -f įor these examples, I need more than a few rows to exist in order to properly examine different data structures. When you have a working installation, you should have an innodb_space command in your path:Įrror: File must be provided with -f argument

#Omnigraffle subscript manual#

If that doesn’t work, you might want to check out The RubyGems manual to try and get your installation working.

If you’re familiar with Ruby and gems (or you just happen to have a well-configured Ruby installation), I regularly push innodb_ruby gems to RubyGems, so you should only need to: We’ll come back to those structures later on! Installing innodb_ruby I won’t try to explain all of the InnoDB structures exposed, since that will get the demos here way off track. Now I’ll show off a few of the things it can do. In On learning InnoDB: A journey to the core I introduced a new library and command-line tool in the innodb_ruby project.