Comments on: How to calculate a good InnoDB log file size

By: AFTAB

AFTAB — Thu, 09 Feb 2012 12:10:56 +0000

When calculating innodb log file space usage; If we consider only “Log sequence number” and doesn’t account for checkpoints (Last checkpoint at) at all. When activity is flushed to the disk – if InnoDB has reached a checkpoint multiple times within the time (60 seconds apart) between status outputs, you might be setting the file size inappropriately??

Anyway this is out of date as of MySQL 5.5, recovery times have been greatly improved and the whole log file flushing algorithm has been improved. In 5.5 you generally want larger log files as recovery is improved. Thus, the large innodb log files allow you to take advantage of the new algorithm.

By: Baron Schwartz

Baron Schwartz — Tue, 01 Jun 2010 14:45:40 +0000

Could you ask this question on our forum instead? http://forum.percona.com/

By: Catalin

Catalin — Tue, 01 Jun 2010 08:56:56 +0000

Hi,

Do you guys know why my innodb data and log files haven’t changed ?
ibdata1 has 1.8G
ib_logfile0 has 5M
ib_logfile1 has 5M

Here are my settings:

innodb_data_home_dir = /var/lib/mysql/
innodb_data_file_path = ibdata1:1000M;ibdata2:10M:autoextend
innodb_log_group_home_dir = /var/lib/mysql/
innodb_log_arch_dir = /var/lib/mysql/
innodb_buffer_pool_size = 2048M
innodb_additional_mem_pool_size = 20M
innodb_log_file_size = 512M
innodb_log_files_in_group = 2
innodb_log_buffer_size = 8M
innodb_thread_concurrency = 8
innodb_flush_log_at_trx_commit = 1
innodb_lock_wait_timeout = 50

I stopped mysqld, deleted the data+log files, but when I start mysql again the log shows:

100601 09:08:30 mysqld started
InnoDB: The first specified data file ./ibdata1 did not exist:
InnoDB: a new database to be created!
100601 9:08:30 InnoDB: Setting file ./ibdata1 size to 10 MB
InnoDB: Database physically writes the file full: wait…
100601 9:08:30 InnoDB: Log file ./ib_logfile0 did not exist: new to be created
InnoDB: Setting log file ./ib_logfile0 size to 5 MB
InnoDB: Database physically writes the file full: wait…
100601 9:08:30 InnoDB: Log file ./ib_logfile1 did not exist: new to be created
InnoDB: Setting log file ./ib_logfile1 size to 5 MB
InnoDB: Database physically writes the file full: wait…
InnoDB: Doublewrite buffer not found: creating new
InnoDB: Doublewrite buffer created
InnoDB: Creating foreign key constraint system tables
InnoDB: Foreign key constraint system tables created
100601 9:08:30 InnoDB: Started; log sequence number 0 0
100601 9:08:30 [Note] /usr/libexec/mysqld: ready for connections.
Version: ’5.0.77′ socket: ‘/var/lib/mysql/mysql.sock’ port: 3306 Source distribution
A mysqld process already exists at Tue Jun 1 09:21:35 BST 2010
100601 9:39:04 InnoDB: ERROR: the age of the last checkpoint is 9434190,
InnoDB: which exceeds the log group capacity 9433498.
InnoDB: If you are using big BLOB or TEXT rows, you must set the
InnoDB: combined size of log files at least 10 times bigger than the
InnoDB: largest such row.

By: Baron Schwartz

Baron Schwartz — Mon, 20 Apr 2009 22:28:05 +0000

Yes, I mean innodb_log_file_size. About the slides — actually that advice was from Peter Zaitsev. He’s not giving you all the details you need to understand what he was talking about. You would have had to be in that talk to understand it.

By: Andy Geers

Andy Geers — Mon, 20 Apr 2009 11:38:28 +0000

I’m slightly confused by this article – it seems to match very closely the advice given by the InnoDB developers for the innodb_log_buffer_size setting here: http://www.innodb.com/wp/wp-content/uploads/2007/05/uc2007-innodb-performance-optimization.pdf

… rather than the innodb_log_*file*_size setting that you are talking about. Can you confirm that you definitely mean the setting you’re referring to?

By: John Marc

John Marc — Sat, 28 Mar 2009 20:55:11 +0000

Yow! Guess I need to up this!

-rw-rw—- 1 mysql mysql 100M Mar 28 16:49 ib_logfile0
-rw-rw—- 1 mysql mysql 100M Mar 28 16:50 ib_logfile1

Using Barons calcs I am looking at 200MB, and Sheens method says 600MB?

By: Vladimir Rusinov

Vladimir Rusinov — Sun, 01 Feb 2009 23:31:43 +0000

Hi. Here is translation to Russian (if someone interested): http://greenmice.info/ru/node/92

By: Baron Schwartz

Baron Schwartz — Sun, 04 Jan 2009 21:37:00 +0000

That is exactly the point of writing this post. Those guidelines are wrong. Both of them.

By: Adam

Adam — Sun, 04 Jan 2009 20:21:55 +0000

Heh, I am messing with this now for my research and I’m quite certain it says right in the readme that it should be 25% of your buffer pool size, which should be 50-80% of available memory. To me it seems like 64M is indeed way too low.

By: Jesper Wisborg Krogh

Jesper Wisborg Krogh — Tue, 23 Dec 2008 04:25:26 +0000

Thanks a lot – I had considered that as an option as well as well as keeping track of which rows actually need to be changed. It will of course require a bit more work and bookkeeping, but it should be doable and I believe in this case it is the best long term solution.

Merry Christmas