pt-index-usage - Read queries from a log and analyze how they use indexes.
pt-index-usage [OPTION...] [FILE...]
pt-index-usage reads queries from logs and analyzes how they use indexes.
Analyze queries in slow.log and print reports:
pt-index-usage /path/to/slow.log --host localhost
Disable reports and save results to mk database for later analysis:
pt-index-usage slow.log --no-report --save-results-database mk
The following section is included to inform users about the potential risks, whether known or unknown, of using this tool. The two main categories of risks are those created by the nature of the tool (e.g. read-only tools vs. read-write tools) and those created by bugs.
This tool is read-only unless you use --save-results-database. It reads a log of queries and EXPLAIN them. It also gathers information about all tables in all databases. It should be very low-risk.
At the time of this release, we know of no bugs that could cause serious harm to users.
The authoritative source for updated information is always the online issue tracking system. Issues that affect this tool will be marked as such. You can see a list of such issues at the following URL: http://www.percona.com/bugs/pt-index-usage.
See also “BUGS” for more information on filing bugs and getting help.
This tool connects to a MySQL database server, reads through a query log, and uses EXPLAIN to ask MySQL how it will use each query. When it is finished, it prints out a report on indexes that the queries didn’t use.
The query log needs to be in MySQL’s slow query log format. If you need to input a different format, you can use pt-query-digest to translate the formats. If you don’t specify a filename, the tool reads from STDIN.
The tool runs two stages. In the first stage, the tool takes inventory of all the tables and indexes in your database, so it can compare the existing indexes to those that were actually used by the queries in the log. In the second stage, it runs EXPLAIN on each query in the query log. It uses separate database connections to inventory the tables and run EXPLAIN, so it opens two connections to the database.
If a query is not a SELECT, it tries to transform it to a roughly equivalent SELECT query so it can be EXPLAINed. This is not a perfect process, but it is good enough to be useful.
The tool skips the EXPLAIN step for queries that are exact duplicates of those seen before. It assumes that the same query will generate the same EXPLAIN plan as it did previously (usually a safe assumption, and generally good for performance), and simply increments the count of times that the indexes were used. However, queries that have the same fingerprint but different checksums will be re-EXPLAINed. Queries that have different literal constants can have different execution plans, and this is important to measure.
After EXPLAIN-ing the query, it is necessary to try to map aliases in the query back to the original table names. For example, consider the EXPLAIN plan for the following query:
SELECT * FROM tbl1 AS foo;
The EXPLAIN output will show access to table foo, and that must be translated back to tbl1. This process involves complex parsing. It is generally very accurate, but there is some chance that it might not work right. If you find cases where it fails, submit a bug report and a reproducible test case.
Queries that cannot be EXPLAINed will cause all subsequent queries with the same fingerprint to be blacklisted. This is to reduce the work they cause, and prevent them from continuing to print error messages. However, at least in this stage of the tool’s development, it is my opinion that it’s not a good idea to preemptively silence these, or prevent them from being EXPLAINed at all. I am looking for lots of feedback on how to improve things like the query parsing. So please submit your test cases based on the errors the tool prints!
After it reads all the events in the log, the tool prints out DROP statements for every index that was not used. It skips indexes for tables that were never accessed by any queries in the log, to avoid false-positive results.
If you don’t specify --quiet, the tool also outputs warnings about statements that cannot be EXPLAINed and similar. These go to standard error.
Progress reports are enabled by default (see --progress). These also go to standard error.
This tool accepts additional command-line arguments. Refer to the “SYNOPSIS” and usage information for details.
Prompt for a password when connecting to MySQL.
short form: -A; type: string
Default character set. If the value is utf8, sets Perl’s binmode on STDOUT to utf8, passes the mysql_enable_utf8 option to DBD::mysql, and runs SET NAMES UTF8 after connecting to MySQL. Any other value sets binmode on STDOUT without the utf8 layer, and runs SET NAMES after connecting to MySQL.
Read this comma-separated list of config files; if specified, this must be the first option on the command line.
Create the --save-results-database if it does not exist.
If the --save-results-database already exists and this option is specified, the database is used and the necessary tables are created if they do not already exist.
Create views for --save-results-database example queries.
Several example queries are given for querying the tables in the --save-results-database. These example queries are, by default, created as views. Specifying --no-create-views prevents these views from being created.
short form: -D; type: string
The database to use for the connection.
short form: -d; type: hash
Only get tables and indexes from this comma-separated list of databases.
Only get tables and indexes from database whose names match this Perl regex.
short form: -F; type: string
Only read mysql options from the given file. You must give an absolute pathname.
type: Hash; default: non-unique
Suggest dropping only these types of unused indexes.
By default pt-index-usage will only suggest to drop unused secondary indexes, not primary or unique indexes. You can specify which types of unused indexes the tool suggests to drop: primary, unique, non-unique, all.
A separate ALTER TABLE statement for each type is printed. So if you specify --drop all and there is a primary key and a non-unique index, the ALTER TABLE ... DROP for each will be printed on separate lines.
Drop and re-create all pre-existing tables in the --save-results-database. This allows information from previous runs to be removed before the current run.
Show help and exit.
short form: -h; type: string
Connect to host.
Ignore this comma-separated list of databases.
Ignore databases whose names match this Perl regex.
Ignore this comma-separated list of table names.
Table names may be qualified with the database name.
Ignore tables whose names match the Perl regex.
short form: -p; type: string
Password to use when connecting.
short form: -P; type: int
Port number to use for connection.
type: array; default: time,30
Print progress reports to STDERR. The value is a comma-separated list with two parts. The first part can be percentage, time, or iterations; the second part specifies how often an update should be printed, in percentage, seconds, or number of iterations.
Print the reports for --report-format.
You may want to disable the reports by specifying --no-report if, for example, you also specify --save-results-database and you only want to query the results tables later.
type: Array; default: drop_unused_indexes
Right now there is only one report: drop_unused_indexes. This report prints SQL statements for dropping any unused indexes. See also --drop.
See also --[no]report.
Save results to tables in this database. Information about indexes, queries, tables and their usage is stored in several tables in the specified database. The tables are auto-created if they do not exist. If the database doesn’t exist, it can be auto-created with --create-save-results-database. In this case the connection is initially created with no default database, then after the database is created, it is USE’ed.
pt-index-usage executes INSERT statements to save the results. Therefore, you should be careful if you use this feature on a production server. It might increase load, or cause trouble if you don’t want the server to be written to, or so on.
This is a new feature. It may change in future releases.
After a run, you can query the usage tables to answer various questions about index usage. The tables have the following CREATE TABLE definitions:
CREATE TABLE IF NOT EXISTS indexes ( db VARCHAR(64) NOT NULL, tbl VARCHAR(64) NOT NULL, idx VARCHAR(64) NOT NULL, cnt BIGINT UNSIGNED NOT NULL DEFAULT 0, PRIMARY KEY (db, tbl, idx) )
CREATE TABLE IF NOT EXISTS queries ( query_id BIGINT UNSIGNED NOT NULL, fingerprint TEXT NOT NULL, sample TEXT NOT NULL, PRIMARY KEY (query_id) )
CREATE TABLE IF NOT EXISTS tables ( db VARCHAR(64) NOT NULL, tbl VARCHAR(64) NOT NULL, cnt BIGINT UNSIGNED NOT NULL DEFAULT 0, PRIMARY KEY (db, tbl) )
CREATE TABLE IF NOT EXISTS index_usage ( query_id BIGINT UNSIGNED NOT NULL, db VARCHAR(64) NOT NULL, tbl VARCHAR(64) NOT NULL, idx VARCHAR(64) NOT NULL, cnt BIGINT UNSIGNED NOT NULL DEFAULT 1, UNIQUE INDEX (query_id, db, tbl, idx) )
CREATE TABLE IF NOT EXISTS index_alternatives ( query_id BIGINT UNSIGNED NOT NULL, -- This query used db VARCHAR(64) NOT NULL, -- this index, but... tbl VARCHAR(64) NOT NULL, -- idx VARCHAR(64) NOT NULL, -- alt_idx VARCHAR(64) NOT NULL, -- was an alternative cnt BIGINT UNSIGNED NOT NULL DEFAULT 1, UNIQUE INDEX (query_id, db, tbl, idx, alt_idx), INDEX (db, tbl, idx), INDEX (db, tbl, alt_idx) )
The following are some queries you can run against these tables to answer common questions you might have. Each query is also created as a view (with MySQL v5.0 and newer) if :option:`--[no]create-views` is true (it is by default). The view names are the strings after the MAGIC_view_ prefix.
Question: which queries sometimes use different indexes, and what fraction of the time is each index chosen? MAGIC_view_query_uses_several_indexes:
SELECT iu.query_id, CONCAT_WS('.', iu.db, iu.tbl, iu.idx) AS idx, variations, iu.cnt, iu.cnt / total_cnt * 100 AS pct FROM index_usage AS iu INNER JOIN ( SELECT query_id, db, tbl, SUM(cnt) AS total_cnt, COUNT(*) AS variations FROM index_usage GROUP BY query_id, db, tbl HAVING COUNT(*) > 1 ) AS qv USING(query_id, db, tbl);
Question: which indexes have lots of alternatives, i.e. are chosen instead of other indexes, and for what queries? MAGIC_view_index_has_alternates:
SELECT CONCAT_WS('.', db, tbl, idx) AS idx_chosen, GROUP_CONCAT(DISTINCT alt_idx) AS alternatives, GROUP_CONCAT(DISTINCT query_id) AS queries, SUM(cnt) AS cnt FROM index_alternatives GROUP BY db, tbl, idx HAVING COUNT(*) > 1;
Question: which indexes are considered as alternates for other indexes, and for what queries? MAGIC_view_index_alternates:
SELECT CONCAT_WS('.', db, tbl, alt_idx) AS idx_considered, GROUP_CONCAT(DISTINCT idx) AS alternative_to, GROUP_CONCAT(DISTINCT query_id) AS queries, SUM(cnt) AS cnt FROM index_alternatives GROUP BY db, tbl, alt_idx HAVING COUNT(*) > 1;
Question: which of those are never chosen by any queries, and are therefore superfluous? MAGIC_view_unused_index_alternates:
SELECT CONCAT_WS('.', i.db, i.tbl, i.idx) AS idx, alt.alternative_to, alt.queries, alt.cnt FROM indexes AS i INNER JOIN ( SELECT db, tbl, alt_idx, GROUP_CONCAT(DISTINCT idx) AS alternative_to, GROUP_CONCAT(DISTINCT query_id) AS queries, SUM(cnt) AS cnt FROM index_alternatives GROUP BY db, tbl, alt_idx HAVING COUNT(*) > 1 ) AS alt ON i.db = alt.db AND i.tbl = alt.tbl AND i.idx = alt.alt_idx WHERE i.cnt = 0;
Question: given a table, which indexes were used, by how many queries, with how many distinct fingerprints? Were there alternatives? Which indexes were not used? You can edit the following query’s SELECT list to also see the query IDs in question. MAGIC_view_index_usage:
SELECT i.idx, iu.usage_cnt, iu.usage_total, ia.alt_cnt, ia.alt_total FROM indexes AS i LEFT OUTER JOIN ( SELECT db, tbl, idx, COUNT(*) AS usage_cnt, SUM(cnt) AS usage_total, GROUP_CONCAT(query_id) AS used_by FROM index_usage GROUP BY db, tbl, idx ) AS iu ON i.db=iu.db AND i.tbl=iu.tbl AND i.idx = iu.idx LEFT OUTER JOIN ( SELECT db, tbl, idx, COUNT(*) AS alt_cnt, SUM(cnt) AS alt_total, GROUP_CONCAT(query_id) AS alt_queries FROM index_alternatives GROUP BY db, tbl, idx ) AS ia ON i.db=ia.db AND i.tbl=ia.tbl AND i.idx = ia.idx;
Question: which indexes on a given table are vital for at least one query (there is no alternative)? MAGIC_view_required_indexes:
SELECT i.db, i.tbl, i.idx, no_alt.queries FROM indexes AS i INNER JOIN ( SELECT iu.db, iu.tbl, iu.idx, GROUP_CONCAT(iu.query_id) AS queries FROM index_usage AS iu LEFT OUTER JOIN index_alternatives AS ia USING(db, tbl, idx) WHERE ia.db IS NULL GROUP BY iu.db, iu.tbl, iu.idx ) AS no_alt ON no_alt.db = i.db AND no_alt.tbl = i.tbl AND no_alt.idx = i.idx ORDER BY i.db, i.tbl, i.idx, no_alt.queries;
type: string; default: wait_timeout=10000
Set these MySQL variables. Immediately after connecting to MySQL, this string will be appended to SET and executed.
short form: -S; type: string
Socket file to use for connection.
short form: -t; type: hash
Only get indexes from this comma-separated list of tables.
Only get indexes from tables whose names match this Perl regex.
short form: -u; type: string
User for login if not current user.
Show version and exit.
These DSN options are used to create a DSN. Each option is given like option=value. The options are case-sensitive, so P and p are not the same option. There cannot be whitespace before or after the = and if the value contains whitespace it must be quoted. DSN options are comma-separated. See the percona-toolkit manpage for full details.
dsn: charset; copy: yes
Default character set.
dsn: database; copy: yes
Database to connect to.
dsn: mysql_read_default_file; copy: yes
Only read default options from the given file
dsn: host; copy: yes
Connect to host.
dsn: password; copy: yes
Password to use when connecting.
dsn: port; copy: yes
Port number to use for connection.
dsn: mysql_socket; copy: yes
Socket file to use for connection.
dsn: user; copy: yes
User for login if not current user.
The environment variable PTDEBUG enables verbose debugging output to STDERR. To enable debugging and capture all output to a file, run the tool like:
PTDEBUG=1 pt-index-usage ... > FILE 2>&1
Be careful: debugging output is voluminous and can generate several megabytes of output.
You need Perl, DBI, DBD::mysql, and some core packages that ought to be installed in any reasonably new version of Perl.
For a list of known bugs, see http://www.percona.com/bugs/pt-index-usage.
Please report bugs at https://bugs.launchpad.net/percona-toolkit. Include the following information in your bug report:
If possible, include debugging output by running the tool with PTDEBUG; see “ENVIRONMENT”.
Visit http://www.percona.com/software/percona-toolkit/ to download the latest release of Percona Toolkit. Or, get the latest release from the command line:
wget percona.com/get/percona-toolkit.tar.gz wget percona.com/get/percona-toolkit.rpm wget percona.com/get/percona-toolkit.deb
You can also get individual tools from the latest release:
Replace TOOL with the name of any tool.
This tool is part of Percona Toolkit, a collection of advanced command-line tools developed by Percona for MySQL support and consulting. Percona Toolkit was forked from two projects in June, 2011: Maatkit and Aspersa. Those projects were created by Baron Schwartz and developed primarily by him and Daniel Nichter, both of whom are employed by Percona. Visit http://www.percona.com/software/ for more software developed by Percona.
This program is copyright 2010-2011 Baron Schwartz, 2011-2012 Percona Inc. Feedback and improvements are welcome.
THIS PROGRAM IS PROVIDED “AS IS” AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 2; OR the Perl Artistic License. On UNIX and similar systems, you can issue `man perlgpl’ or `man perlartistic’ to read these licenses.
You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.