Among the many things that can cause a “server stall” is a long-running transaction. If a transaction remains open for a very long time without committing, and has modified data, then other transactions could block and fail with a lock wait timeout. The problem is, it can be very difficult to find the offending code [...]
Advanced index analysis with mk-index-usage
November 11, 2010 By 2 Comments
The new release of Maatkit has a useful feature in mk-index-usage to help you determine how indexes are used in more flexible ways. The default report just prints out ALTER statements for removing unused indexes, which is nice, but it’s often helpful to ask more sophisticated questions about index usage. I’ll use this blog’s queries [...]
Lost innodb tables, xfs and binary grep
November 9, 2010 By 12 Comments
Before I start a story about the data recovery case I worked on yesterday, here’s a quick tip – having a database backup does not mean you can restore from it. Always verify your backup can be used to restore the database! If not automatically, do this manually, at least once a month. No, seriously [...]
Two Types of MySQL Users
May 10, 2010 By 5 Comments
In comments to my previous post I got number number of comments saying if MySQL would not have multiple storage engine interface it would not allow people to do various very cool stuff. And I agree with this. The question is how cool you want your database operation to be ? Visiting customers I see [...]
Why you should ignore MySQL’s key cache hit ratio
February 28, 2010 By 28 Comments
I have not caused a fist fight in a while, so it’s time to take off the gloves. I claim that somewhere around of 99% of advice about tuning MySQL’s key cache hit ratio is wrong, even when you hear it from experts. There are two major problems with the key buffer hit ratio, and [...]
How (not) to find unused indexes
October 16, 2009 By 2 Comments
I’ve seen a few people link to an INFORMATION_SCHEMA query to be able to find any indexes that have low cardinality, in an effort to find out what indexes should be removed. This method is flawed – here’s the first reason why:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 | CREATE TABLE `sales` ( `id` int(11) NOT NULL AUTO_INCREMENT, `customer_id` int(11) DEFAULT NULL, `status` enum('archived','active') DEFAULT NULL, PRIMARY KEY (`id`), KEY `status` (`status`) ) ENGINE=MyISAM AUTO_INCREMENT=65691 DEFAULT CHARSET=latin1; mysql> SELECT count(*), status FROM sales GROUP by status; +----------+---------+ | count(*) | status | +----------+---------+ |   65536 | archived | |     154 | active | +----------+---------+ 2 rows in set (0.17 sec) mysql> EXPLAIN SELECT * FROM sales WHERE status='active'; # query 1 +----+-------------+-------+------+---------------+--------+---------+-------+------+-------------+ | id | select_type | table | type | possible_keys | key   | key_len | ref  | rows | Extra      | +----+-------------+-------+------+---------------+--------+---------+-------+------+-------------+ | 1 | SIMPLE     | sales | ref | status       | status | 2      | const | 196 | Using where | +----+-------------+-------+------+---------------+--------+---------+-------+------+-------------+ 1 row in set (0.06 sec) mysql> EXPLAIN SELECT * FROM sales WHERE status='archived'; # query 2 +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra      | +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+ | 1 | SIMPLE     | sales | ALL | status       | NULL | NULL   | NULL | 65690 | Using where | +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+ 1 row in set (0.01 sec) |
The cardinality of status index is woeful, but provided that the application [...]
Quick comparison of MyISAM, Infobright, and MonetDB
September 29, 2009 By 30 Comments
Recently I was doing a little work for a client who has MyISAM tables with many columns (the same one Peter wrote about recently). The client’s performance is suffering in part because of the number of columns, which is over 200. The queries are generally pretty simple (sums of columns), but they’re ad-hoc (can access [...]
Gathering queries from a server with Maatkit and tcpdump
July 1, 2009 By 23 Comments
For the last couple of months, we’ve been quietly developing a MySQL protocol parser for Maatkit. It isn’t an implementation of the protocol: it’s an observer of the protocol. This lets us gather queries from servers that don’t have a slow query log enabled, at very high time resolution. With this new functionality, it becomes [...]
Faster MySQL failover with SELECT mirroring
February 1, 2009 By 9 Comments
One of my favorite MySQL configurations for high availability is master-master replication, which is just like normal master-slave replication except that you can fail over in both directions. Aside from MySQL Cluster, which is more special-purpose, this is probably the best general-purpose way to get fast failover and a bunch of other benefits (non-blocking ALTER [...]
The perils of InnoDB with Debian and startup scripts
January 28, 2009 By 13 Comments
Are you running MySQL on Debian or Ubuntu with InnoDB? You might want to disable /etc/mysql/debian-start. When you run /etc/init.d/mysql start it runs this script, which runs mysqlcheck, which can destroy performance. It can happen on a server with MyISAM tables, if there are enough tables, but it is far worse on InnoDB. There are [...]
