2015-03-11 03:33:24 +01:00
|
|
|
<!-- doc/src/sgml/ref/pgbench.sgml -->
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refentry id="pgbench">
|
2014-02-24 03:25:35 +01:00
|
|
|
<indexterm zone="pgbench">
|
|
|
|
<primary>pgbench</primary>
|
|
|
|
</indexterm>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refmeta>
|
|
|
|
<refentrytitle><application>pgbench</application></refentrytitle>
|
|
|
|
<manvolnum>1</manvolnum>
|
|
|
|
<refmiscinfo>Application</refmiscinfo>
|
|
|
|
</refmeta>
|
|
|
|
|
|
|
|
<refnamediv>
|
|
|
|
<refname>pgbench</refname>
|
|
|
|
<refpurpose>run a benchmark test on <productname>PostgreSQL</productname></refpurpose>
|
|
|
|
</refnamediv>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsynopsisdiv>
|
|
|
|
<cmdsynopsis>
|
|
|
|
<command>pgbench</command>
|
|
|
|
<arg choice="plain"><option>-i</option></arg>
|
|
|
|
<arg rep="repeat"><replaceable>option</replaceable></arg>
|
|
|
|
<arg choice="opt"><replaceable>dbname</replaceable></arg>
|
|
|
|
</cmdsynopsis>
|
|
|
|
<cmdsynopsis>
|
|
|
|
<command>pgbench</command>
|
|
|
|
<arg rep="repeat"><replaceable>option</replaceable></arg>
|
|
|
|
<arg choice="opt"><replaceable>dbname</replaceable></arg>
|
|
|
|
</cmdsynopsis>
|
|
|
|
</refsynopsisdiv>
|
|
|
|
|
|
|
|
<refsect1>
|
|
|
|
<title>Description</title>
|
2010-03-23 05:09:17 +01:00
|
|
|
<para>
|
|
|
|
<application>pgbench</application> is a simple program for running benchmark
|
2017-10-09 03:44:17 +02:00
|
|
|
tests on <productname>PostgreSQL</productname>. It runs the same sequence of SQL
|
2010-03-23 05:09:17 +01:00
|
|
|
commands over and over, possibly in multiple concurrent database sessions,
|
|
|
|
and then calculates the average transaction rate (transactions per second).
|
|
|
|
By default, <application>pgbench</application> tests a scenario that is
|
2017-10-09 03:44:17 +02:00
|
|
|
loosely based on TPC-B, involving five <command>SELECT</command>,
|
|
|
|
<command>UPDATE</command>, and <command>INSERT</command> commands per transaction.
|
2010-03-23 05:09:17 +01:00
|
|
|
However, it is easy to test other cases by writing your own transaction
|
|
|
|
script files.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2015-09-11 03:22:21 +02:00
|
|
|
Typical output from <application>pgbench</application> looks like:
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2010-07-29 21:34:41 +02:00
|
|
|
<screen>
|
2016-01-27 02:54:22 +01:00
|
|
|
transaction type: <builtin: TPC-B (sort of)>
|
2010-03-23 05:09:17 +01:00
|
|
|
scaling factor: 10
|
|
|
|
query mode: simple
|
|
|
|
number of clients: 10
|
|
|
|
number of threads: 1
|
|
|
|
number of transactions per client: 1000
|
|
|
|
number of transactions actually processed: 10000/10000
|
|
|
|
tps = 85.184871 (including connections establishing)
|
|
|
|
tps = 85.296346 (excluding connections establishing)
|
2010-07-29 21:34:41 +02:00
|
|
|
</screen>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
|
|
|
The first six lines report some of the most important parameter
|
|
|
|
settings. The next line reports the number of transactions completed
|
|
|
|
and intended (the latter being just the product of number of clients
|
|
|
|
and number of transactions per client); these will be equal unless the run
|
2017-10-09 03:44:17 +02:00
|
|
|
failed before completion. (In <option>-T</option> mode, only the actual
|
2010-08-12 22:39:39 +02:00
|
|
|
number of transactions is printed.)
|
2010-08-15 21:46:24 +02:00
|
|
|
The last two lines report the number of transactions per second,
|
2010-03-23 05:09:17 +01:00
|
|
|
figured with and without counting the time to start database sessions.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
The default TPC-B-like transaction test requires specific tables to be
|
2017-10-09 03:44:17 +02:00
|
|
|
set up beforehand. <application>pgbench</application> should be invoked with
|
|
|
|
the <option>-i</option> (initialize) option to create and populate these
|
2010-03-23 05:09:17 +01:00
|
|
|
tables. (When you are testing a custom script, you don't need this
|
|
|
|
step, but will instead need to do whatever setup your test needs.)
|
|
|
|
Initialization looks like:
|
|
|
|
|
2010-07-29 21:34:41 +02:00
|
|
|
<programlisting>
|
2017-10-09 03:44:17 +02:00
|
|
|
pgbench -i <optional> <replaceable>other-options</replaceable> </optional> <replaceable>dbname</replaceable>
|
2010-07-29 21:34:41 +02:00
|
|
|
</programlisting>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2017-10-09 03:44:17 +02:00
|
|
|
where <replaceable>dbname</replaceable> is the name of the already-created
|
|
|
|
database to test in. (You may also need <option>-h</option>,
|
|
|
|
<option>-p</option>, and/or <option>-U</option> options to specify how to
|
2010-03-23 05:09:17 +01:00
|
|
|
connect to the database server.)
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<caution>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>pgbench -i</literal> creates four tables <structname>pgbench_accounts</structname>,
|
|
|
|
<structname>pgbench_branches</structname>, <structname>pgbench_history</structname>, and
|
|
|
|
<structname>pgbench_tellers</structname>,
|
2010-03-23 05:09:17 +01:00
|
|
|
destroying any existing tables of these names.
|
|
|
|
Be very careful to use another database if you have tables having these
|
|
|
|
names!
|
|
|
|
</para>
|
|
|
|
</caution>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
At the default <quote>scale factor</quote> of 1, the tables initially
|
2010-03-23 05:09:17 +01:00
|
|
|
contain this many rows:
|
2010-07-29 21:34:41 +02:00
|
|
|
<screen>
|
2010-03-23 05:09:17 +01:00
|
|
|
table # of rows
|
|
|
|
---------------------------------
|
|
|
|
pgbench_branches 1
|
|
|
|
pgbench_tellers 10
|
|
|
|
pgbench_accounts 100000
|
|
|
|
pgbench_history 0
|
2010-07-29 21:34:41 +02:00
|
|
|
</screen>
|
2010-03-23 05:09:17 +01:00
|
|
|
You can (and, for most purposes, probably should) increase the number
|
2017-10-09 03:44:17 +02:00
|
|
|
of rows by using the <option>-s</option> (scale factor) option. The
|
|
|
|
<option>-F</option> (fillfactor) option might also be used at this point.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Once you have done the necessary setup, you can run your benchmark
|
2017-10-09 03:44:17 +02:00
|
|
|
with a command that doesn't include <option>-i</option>, that is
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2010-07-29 21:34:41 +02:00
|
|
|
<programlisting>
|
2017-10-09 03:44:17 +02:00
|
|
|
pgbench <optional> <replaceable>options</replaceable> </optional> <replaceable>dbname</replaceable>
|
2010-07-29 21:34:41 +02:00
|
|
|
</programlisting>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
|
|
|
In nearly all cases, you'll need some options to make a useful test.
|
2017-10-09 03:44:17 +02:00
|
|
|
The most important options are <option>-c</option> (number of clients),
|
|
|
|
<option>-t</option> (number of transactions), <option>-T</option> (time limit),
|
|
|
|
and <option>-f</option> (specify a custom script file).
|
2010-03-23 05:09:17 +01:00
|
|
|
See below for a full list.
|
|
|
|
</para>
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect1>
|
|
|
|
|
|
|
|
<refsect1>
|
|
|
|
<title>Options</title>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
|
|
|
<para>
|
Allow running just selected steps of pgbench's initialization sequence.
This feature caters to specialized use-cases such as running the normal
pgbench scenario with nonstandard indexes, or inserting other actions
between steps of the initialization sequence. The normal sequence of
initialization actions is broken down into half a dozen steps which can
be executed in a user-specified order, to the extent to which that's
sensible. The actions themselves aren't changed, except to make them
more robust against nonstandard uses:
* all four tables are now dropped in one DROP command, to reduce
assumptions about what foreign key relationships exist;
* all four tables are now truncated at the start of the data load
step, for consistency;
* the foreign key creation commands now specify constraint names, to
prevent accidentally creating duplicate constraints by executing the
'f' step twice.
Make some cosmetic adjustments in the messages emitted by pgbench
so that it's clear which steps are getting run, and so that the
messages agree with the documented names of the steps.
In passing, fix failure to enforce that the -v option is used only
in benchmarking mode.
Masahiko Sawada, reviewed by Fabien Coelho, editorialized a bit by me
Discussion: https://postgr.es/m/CAD21AoCsz0ZzfCFcxYZ+PUdpkDd5VsCSG0Pre_-K1EgokCDFYA@mail.gmail.com
2017-11-13 22:40:03 +01:00
|
|
|
The following is divided into three subsections. Different options are
|
|
|
|
used during database initialization and while running benchmarks, but some
|
|
|
|
options are useful in both cases.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2 id="pgbench-init-options">
|
|
|
|
<title>Initialization Options</title>
|
2010-05-25 17:55:28 +02:00
|
|
|
|
|
|
|
<para>
|
|
|
|
<application>pgbench</application> accepts the following command-line
|
|
|
|
initialization arguments:
|
|
|
|
|
|
|
|
<variablelist>
|
|
|
|
|
|
|
|
<varlistentry>
|
2010-08-12 22:39:39 +02:00
|
|
|
<term><option>-i</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--initialize</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2010-08-12 22:39:39 +02:00
|
|
|
Required to invoke initialization mode.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
Allow running just selected steps of pgbench's initialization sequence.
This feature caters to specialized use-cases such as running the normal
pgbench scenario with nonstandard indexes, or inserting other actions
between steps of the initialization sequence. The normal sequence of
initialization actions is broken down into half a dozen steps which can
be executed in a user-specified order, to the extent to which that's
sensible. The actions themselves aren't changed, except to make them
more robust against nonstandard uses:
* all four tables are now dropped in one DROP command, to reduce
assumptions about what foreign key relationships exist;
* all four tables are now truncated at the start of the data load
step, for consistency;
* the foreign key creation commands now specify constraint names, to
prevent accidentally creating duplicate constraints by executing the
'f' step twice.
Make some cosmetic adjustments in the messages emitted by pgbench
so that it's clear which steps are getting run, and so that the
messages agree with the documented names of the steps.
In passing, fix failure to enforce that the -v option is used only
in benchmarking mode.
Masahiko Sawada, reviewed by Fabien Coelho, editorialized a bit by me
Discussion: https://postgr.es/m/CAD21AoCsz0ZzfCFcxYZ+PUdpkDd5VsCSG0Pre_-K1EgokCDFYA@mail.gmail.com
2017-11-13 22:40:03 +01:00
|
|
|
<varlistentry>
|
|
|
|
<term><option>-I <replaceable>init_steps</replaceable></option></term>
|
|
|
|
<term><option>--init-steps=<replaceable>init_steps</replaceable></option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Perform just a selected set of the normal initialization steps.
|
|
|
|
<replaceable>init_steps</replaceable> specifies the
|
|
|
|
initialization steps to be performed, using one character per step.
|
|
|
|
Each step is invoked in the specified order.
|
|
|
|
The default is <literal>dtgvp</literal>.
|
|
|
|
The available steps are:
|
|
|
|
|
|
|
|
<variablelist>
|
|
|
|
<varlistentry>
|
|
|
|
<term><literal>d</literal> (Drop)</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Drop any existing <application>pgbench</application> tables.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
<varlistentry>
|
|
|
|
<term><literal>t</literal> (create Tables)</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create the tables used by the
|
|
|
|
standard <application>pgbench</application> scenario, namely
|
|
|
|
<structname>pgbench_accounts</structname>,
|
|
|
|
<structname>pgbench_branches</structname>,
|
|
|
|
<structname>pgbench_history</structname>, and
|
|
|
|
<structname>pgbench_tellers</structname>.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
<varlistentry>
|
|
|
|
<term><literal>g</literal> (Generate data)</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Generate data and load it into the standard tables,
|
|
|
|
replacing any data already present.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
<varlistentry>
|
|
|
|
<term><literal>v</literal> (Vacuum)</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Invoke <command>VACUUM</command> on the standard tables.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
<varlistentry>
|
|
|
|
<term><literal>p</literal> (create Primary keys)</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create primary key indexes on the standard tables.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
<varlistentry>
|
|
|
|
<term><literal>f</literal> (create Foreign keys)</term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create foreign key constraints between the standard tables.
|
|
|
|
(Note that this step is not performed by default.)
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
</variablelist>
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2013-05-14 03:37:01 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-F</option> <replaceable>fillfactor</replaceable></term>
|
|
|
|
<term><option>--fillfactor=</option><replaceable>fillfactor</replaceable></term>
|
2013-05-14 03:37:01 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Create the <structname>pgbench_accounts</structname>,
|
|
|
|
<structname>pgbench_tellers</structname> and
|
|
|
|
<structname>pgbench_branches</structname> tables with the given fillfactor.
|
2013-05-14 03:37:01 +02:00
|
|
|
Default is 100.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2012-07-23 20:38:34 +02:00
|
|
|
<varlistentry>
|
|
|
|
<term><option>-n</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--no-vacuum</option></term>
|
2012-07-23 20:38:34 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
Allow running just selected steps of pgbench's initialization sequence.
This feature caters to specialized use-cases such as running the normal
pgbench scenario with nonstandard indexes, or inserting other actions
between steps of the initialization sequence. The normal sequence of
initialization actions is broken down into half a dozen steps which can
be executed in a user-specified order, to the extent to which that's
sensible. The actions themselves aren't changed, except to make them
more robust against nonstandard uses:
* all four tables are now dropped in one DROP command, to reduce
assumptions about what foreign key relationships exist;
* all four tables are now truncated at the start of the data load
step, for consistency;
* the foreign key creation commands now specify constraint names, to
prevent accidentally creating duplicate constraints by executing the
'f' step twice.
Make some cosmetic adjustments in the messages emitted by pgbench
so that it's clear which steps are getting run, and so that the
messages agree with the documented names of the steps.
In passing, fix failure to enforce that the -v option is used only
in benchmarking mode.
Masahiko Sawada, reviewed by Fabien Coelho, editorialized a bit by me
Discussion: https://postgr.es/m/CAD21AoCsz0ZzfCFcxYZ+PUdpkDd5VsCSG0Pre_-K1EgokCDFYA@mail.gmail.com
2017-11-13 22:40:03 +01:00
|
|
|
Perform no vacuuming during initialization.
|
|
|
|
(This option suppresses the <literal>v</literal> initialization step,
|
|
|
|
even if it was specified in <option>-I</option>.)
|
2012-07-23 20:38:34 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2010-05-25 17:55:28 +02:00
|
|
|
<varlistentry>
|
2013-05-14 03:37:01 +02:00
|
|
|
<term><option>-q</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--quiet</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2013-05-14 03:37:01 +02:00
|
|
|
Switch logging to quiet mode, producing only one progress message per 5
|
|
|
|
seconds. The default logging prints one message each 100000 rows, which
|
|
|
|
often outputs many lines per second (especially on good hardware).
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-s</option> <replaceable>scale_factor</replaceable></term>
|
|
|
|
<term><option>--scale=</option><replaceable>scale_factor</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Multiply the number of rows generated by the scale factor.
|
2017-10-09 03:44:17 +02:00
|
|
|
For example, <literal>-s 100</literal> will create 10,000,000 rows
|
|
|
|
in the <structname>pgbench_accounts</structname> table. Default is 1.
|
2013-01-29 10:49:40 +01:00
|
|
|
When the scale is 20,000 or larger, the columns used to
|
|
|
|
hold account identifiers (<structfield>aid</structfield> columns)
|
|
|
|
will switch to using larger integers (<type>bigint</type>),
|
|
|
|
in order to be big enough to hold the range of account
|
|
|
|
identifiers.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2012-06-20 00:33:59 +02:00
|
|
|
<varlistentry>
|
|
|
|
<term><option>--foreign-keys</option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create foreign key constraints between the standard tables.
|
Allow running just selected steps of pgbench's initialization sequence.
This feature caters to specialized use-cases such as running the normal
pgbench scenario with nonstandard indexes, or inserting other actions
between steps of the initialization sequence. The normal sequence of
initialization actions is broken down into half a dozen steps which can
be executed in a user-specified order, to the extent to which that's
sensible. The actions themselves aren't changed, except to make them
more robust against nonstandard uses:
* all four tables are now dropped in one DROP command, to reduce
assumptions about what foreign key relationships exist;
* all four tables are now truncated at the start of the data load
step, for consistency;
* the foreign key creation commands now specify constraint names, to
prevent accidentally creating duplicate constraints by executing the
'f' step twice.
Make some cosmetic adjustments in the messages emitted by pgbench
so that it's clear which steps are getting run, and so that the
messages agree with the documented names of the steps.
In passing, fix failure to enforce that the -v option is used only
in benchmarking mode.
Masahiko Sawada, reviewed by Fabien Coelho, editorialized a bit by me
Discussion: https://postgr.es/m/CAD21AoCsz0ZzfCFcxYZ+PUdpkDd5VsCSG0Pre_-K1EgokCDFYA@mail.gmail.com
2017-11-13 22:40:03 +01:00
|
|
|
(This option adds the <literal>f</literal> step to the initialization
|
|
|
|
step sequence, if it is not already present.)
|
2012-06-20 00:33:59 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2011-07-25 15:16:14 +02:00
|
|
|
<varlistentry>
|
|
|
|
<term><option>--index-tablespace=<replaceable>index_tablespace</replaceable></option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create indexes in the specified tablespace, rather than the default
|
|
|
|
tablespace.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>--tablespace=<replaceable>tablespace</replaceable></option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create tables in the specified tablespace, rather than the default
|
|
|
|
tablespace.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2011-07-25 12:49:00 +02:00
|
|
|
<varlistentry>
|
|
|
|
<term><option>--unlogged-tables</option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Create all tables as unlogged tables, rather than permanent tables.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2010-05-25 17:55:28 +02:00
|
|
|
</variablelist>
|
|
|
|
</para>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
2010-05-25 17:55:28 +02:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2 id="pgbench-run-options">
|
|
|
|
<title>Benchmarking Options</title>
|
2010-05-25 17:55:28 +02:00
|
|
|
|
|
|
|
<para>
|
|
|
|
<application>pgbench</application> accepts the following command-line
|
|
|
|
benchmarking arguments:
|
|
|
|
|
|
|
|
<variablelist>
|
2016-01-27 02:54:22 +01:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-b</option> <replaceable>scriptname[@weight]</replaceable></term>
|
|
|
|
<term><option>--builtin</option>=<replaceable>scriptname[@weight]</replaceable></term>
|
2016-01-27 02:54:22 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2016-07-29 04:46:15 +02:00
|
|
|
Add the specified built-in script to the list of executed scripts.
|
2017-10-09 03:44:17 +02:00
|
|
|
An optional integer weight after <literal>@</literal> allows to adjust the
|
pgbench: Allow changing weights for scripts
Previously, all scripts had the same probability of being chosen when
multiple of them were specified via -b, -f, -N, -S. With this commit,
-b and -f now search for an "@" in the script name and use the integer
found after it as the drawing probability for that script.
(One disadvantage is that if you have script whose names contain @, you
are now forced to specify "@1" at the end; otherwise the name's @ is
confused with a weight separator. We don't expect many pgbench script
with @ in their names in the wild, so this shouldn't be too serious a
problem.)
While at it, rework the interface between addScript, process_file,
process_builtin, and findBuiltin. It had gotten a bit out of hand with
recent commits.
Author: Fabien Coelho
Reviewed-By: Andres Freund, Robert Haas, Álvaro Herrera, Michaël Paquier
Discussion: http://www.postgresql.org/message-id/alpine.DEB.2.10.1603160721240.1666@sto
2016-03-19 16:32:42 +01:00
|
|
|
probability of drawing the script. If not specified, it is set to 1.
|
2017-10-09 03:44:17 +02:00
|
|
|
Available built-in scripts are: <literal>tpcb-like</literal>,
|
|
|
|
<literal>simple-update</literal> and <literal>select-only</literal>.
|
2016-07-29 04:46:15 +02:00
|
|
|
Unambiguous prefixes of built-in names are accepted.
|
2017-10-09 03:44:17 +02:00
|
|
|
With special name <literal>list</literal>, show the list of built-in scripts
|
2016-01-27 02:54:22 +01:00
|
|
|
and exit immediately.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2010-05-25 17:55:28 +02:00
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-c</option> <replaceable>clients</replaceable></term>
|
|
|
|
<term><option>--client=</option><replaceable>clients</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Number of clients simulated, that is, number of concurrent database
|
|
|
|
sessions. Default is 1.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-C</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--connect</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Establish a new connection for each transaction, rather than
|
|
|
|
doing it just once per client session.
|
|
|
|
This is useful to measure the connection overhead.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-d</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--debug</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Print debugging output.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-D</option> <replaceable>varname</replaceable><literal>=</literal><replaceable>value</replaceable></term>
|
|
|
|
<term><option>--define=</option><replaceable>varname</replaceable><literal>=</literal><replaceable>value</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Define a variable for use by a custom script (see below).
|
2017-10-09 03:44:17 +02:00
|
|
|
Multiple <option>-D</option> options are allowed.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-f</option> <replaceable>filename[@weight]</replaceable></term>
|
|
|
|
<term><option>--file=</option><replaceable>filename[@weight]</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Add a transaction script read from <replaceable>filename</replaceable> to
|
2016-01-27 02:54:22 +01:00
|
|
|
the list of executed scripts.
|
2017-10-09 03:44:17 +02:00
|
|
|
An optional integer weight after <literal>@</literal> allows to adjust the
|
pgbench: Allow changing weights for scripts
Previously, all scripts had the same probability of being chosen when
multiple of them were specified via -b, -f, -N, -S. With this commit,
-b and -f now search for an "@" in the script name and use the integer
found after it as the drawing probability for that script.
(One disadvantage is that if you have script whose names contain @, you
are now forced to specify "@1" at the end; otherwise the name's @ is
confused with a weight separator. We don't expect many pgbench script
with @ in their names in the wild, so this shouldn't be too serious a
problem.)
While at it, rework the interface between addScript, process_file,
process_builtin, and findBuiltin. It had gotten a bit out of hand with
recent commits.
Author: Fabien Coelho
Reviewed-By: Andres Freund, Robert Haas, Álvaro Herrera, Michaël Paquier
Discussion: http://www.postgresql.org/message-id/alpine.DEB.2.10.1603160721240.1666@sto
2016-03-19 16:32:42 +01:00
|
|
|
probability of drawing the test.
|
2010-05-25 17:55:28 +02:00
|
|
|
See below for details.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-j</option> <replaceable>threads</replaceable></term>
|
|
|
|
<term><option>--jobs=</option><replaceable>threads</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Number of worker threads within <application>pgbench</application>.
|
|
|
|
Using more than one thread can be helpful on multi-CPU machines.
|
2015-07-03 09:45:40 +02:00
|
|
|
Clients are distributed as evenly as possible among available threads.
|
2010-05-25 17:55:28 +02:00
|
|
|
Default is 1.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-l</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--log</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-01-02 18:26:03 +01:00
|
|
|
Write information about each transaction to a log file.
|
2010-05-25 17:55:28 +02:00
|
|
|
See below for details.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2014-10-13 19:25:56 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-L</option> <replaceable>limit</replaceable></term>
|
|
|
|
<term><option>--latency-limit=</option><replaceable>limit</replaceable></term>
|
2014-10-13 19:25:56 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Transaction which last more than <replaceable>limit</replaceable> milliseconds
|
|
|
|
are counted and reported separately, as <firstterm>late</firstterm>.
|
2014-10-13 19:25:56 +02:00
|
|
|
</para>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
When throttling is used (<option>--rate=...</option>), transactions that
|
|
|
|
lag behind schedule by more than <replaceable>limit</replaceable> ms, and thus
|
2014-10-13 19:25:56 +02:00
|
|
|
have no hope of meeting the latency limit, are not sent to the server
|
|
|
|
at all. They are counted and reported separately as
|
2017-10-09 03:44:17 +02:00
|
|
|
<firstterm>skipped</firstterm>.
|
2014-10-13 19:25:56 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2010-05-25 17:55:28 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-M</option> <replaceable>querymode</replaceable></term>
|
|
|
|
<term><option>--protocol=</option><replaceable>querymode</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Protocol to use for submitting queries to the server:
|
|
|
|
<itemizedlist>
|
|
|
|
<listitem>
|
2017-10-09 03:44:17 +02:00
|
|
|
<para><literal>simple</literal>: use simple query protocol.</para>
|
2010-05-25 17:55:28 +02:00
|
|
|
</listitem>
|
|
|
|
<listitem>
|
2017-10-09 03:44:17 +02:00
|
|
|
<para><literal>extended</literal>: use extended query protocol.</para>
|
2010-05-25 17:55:28 +02:00
|
|
|
</listitem>
|
|
|
|
<listitem>
|
2017-10-09 03:44:17 +02:00
|
|
|
<para><literal>prepared</literal>: use extended query protocol with prepared statements.</para>
|
2010-05-25 17:55:28 +02:00
|
|
|
</listitem>
|
|
|
|
</itemizedlist>
|
2017-11-23 15:39:47 +01:00
|
|
|
The default is simple query protocol. (See <xref linkend="protocol"/>
|
2010-05-25 17:55:28 +02:00
|
|
|
for more information.)
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-n</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--no-vacuum</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Perform no vacuuming before running the test.
|
2017-10-09 03:44:17 +02:00
|
|
|
This option is <emphasis>necessary</emphasis>
|
2010-05-25 17:55:28 +02:00
|
|
|
if you are running a custom test scenario that does not include
|
2017-10-09 03:44:17 +02:00
|
|
|
the standard tables <structname>pgbench_accounts</structname>,
|
|
|
|
<structname>pgbench_branches</structname>, <structname>pgbench_history</structname>, and
|
|
|
|
<structname>pgbench_tellers</structname>.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-N</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--skip-some-updates</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2016-07-29 04:46:15 +02:00
|
|
|
Run built-in simple-update script.
|
2017-10-09 03:44:17 +02:00
|
|
|
Shorthand for <option>-b simple-update</option>.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2013-07-17 01:05:37 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-P</option> <replaceable>sec</replaceable></term>
|
|
|
|
<term><option>--progress=</option><replaceable>sec</replaceable></term>
|
2013-07-17 01:05:37 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Show progress report every <replaceable>sec</replaceable> seconds. The report
|
2013-10-05 23:33:38 +02:00
|
|
|
includes the time since the beginning of the run, the tps since the
|
|
|
|
last report, and the transaction latency average and standard
|
2017-10-09 03:44:17 +02:00
|
|
|
deviation since the last report. Under throttling (<option>-R</option>),
|
2014-09-11 11:40:01 +02:00
|
|
|
the latency is computed with respect to the transaction scheduled
|
|
|
|
start time, not the actual transaction beginning time, thus it also
|
|
|
|
includes the average schedule lag time.
|
2013-07-17 01:05:37 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2014-07-15 20:34:33 +02:00
|
|
|
<varlistentry>
|
|
|
|
<term><option>-r</option></term>
|
|
|
|
<term><option>--report-latencies</option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Report the average per-statement latency (execution time from the
|
|
|
|
perspective of the client) of each command after the benchmark
|
|
|
|
finishes. See below for details.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2013-07-23 01:40:22 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-R</option> <replaceable>rate</replaceable></term>
|
|
|
|
<term><option>--rate=</option><replaceable>rate</replaceable></term>
|
2013-07-23 01:40:22 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Execute transactions targeting the specified rate instead of running
|
|
|
|
as fast as possible (the default). The rate is given in transactions
|
|
|
|
per second. If the targeted rate is above the maximum possible rate,
|
|
|
|
the rate limit won't impact the results.
|
|
|
|
</para>
|
|
|
|
<para>
|
|
|
|
The rate is targeted by starting transactions along a
|
2013-09-23 21:49:21 +02:00
|
|
|
Poisson-distributed schedule time line. The expected start time
|
2013-07-23 01:40:22 +02:00
|
|
|
schedule moves forward based on when the client first started, not
|
|
|
|
when the previous transaction ended. That approach means that when
|
|
|
|
transactions go past their original scheduled end time, it is
|
|
|
|
possible for later ones to catch up again.
|
|
|
|
</para>
|
2013-11-10 15:20:52 +01:00
|
|
|
<para>
|
2014-09-11 11:40:01 +02:00
|
|
|
When throttling is active, the transaction latency reported at the
|
|
|
|
end of the run is calculated from the scheduled start times, so it
|
|
|
|
includes the time each transaction had to wait for the previous
|
|
|
|
transaction to finish. The wait time is called the schedule lag time,
|
|
|
|
and its average and maximum are also reported separately. The
|
|
|
|
transaction latency with respect to the actual transaction start time,
|
|
|
|
i.e. the time spent executing the transaction in the database, can be
|
|
|
|
computed by subtracting the schedule lag time from the reported
|
|
|
|
latency.
|
2013-07-23 01:40:22 +02:00
|
|
|
</para>
|
2014-09-11 11:40:01 +02:00
|
|
|
|
2014-10-13 19:25:56 +02:00
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
If <option>--latency-limit</option> is used together with <option>--rate</option>,
|
2014-10-13 19:25:56 +02:00
|
|
|
a transaction can lag behind so much that it is already over the
|
|
|
|
latency limit when the previous transaction ends, because the latency
|
|
|
|
is calculated from the scheduled start time. Such transactions are
|
|
|
|
not sent to the server, but are skipped altogether and counted
|
|
|
|
separately.
|
|
|
|
</para>
|
|
|
|
|
2013-07-23 01:40:22 +02:00
|
|
|
<para>
|
2014-09-11 11:40:01 +02:00
|
|
|
A high schedule lag time is an indication that the system cannot
|
|
|
|
process transactions at the specified rate, with the chosen number of
|
|
|
|
clients and threads. When the average transaction execution time is
|
|
|
|
longer than the scheduled interval between each transaction, each
|
|
|
|
successive transaction will fall further behind, and the schedule lag
|
|
|
|
time will keep increasing the longer the test run is. When that
|
|
|
|
happens, you will have to reduce the specified transaction rate.
|
2013-07-23 01:40:22 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2010-05-25 17:55:28 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-s</option> <replaceable>scale_factor</replaceable></term>
|
|
|
|
<term><option>--scale=</option><replaceable>scale_factor</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Report the specified scale factor in <application>pgbench</application>'s
|
2010-05-25 17:55:28 +02:00
|
|
|
output. With the built-in tests, this is not necessary; the
|
|
|
|
correct scale factor will be detected by counting the number of
|
2017-10-09 03:44:17 +02:00
|
|
|
rows in the <structname>pgbench_branches</structname> table.
|
|
|
|
However, when testing only custom benchmarks (<option>-f</option> option),
|
2016-01-27 02:54:22 +01:00
|
|
|
the scale factor will be reported as 1 unless this option is used.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-S</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--select-only</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2016-01-27 02:54:22 +01:00
|
|
|
Run built-in select-only script.
|
2017-10-09 03:44:17 +02:00
|
|
|
Shorthand for <option>-b select-only</option>.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-t</option> <replaceable>transactions</replaceable></term>
|
|
|
|
<term><option>--transactions=</option><replaceable>transactions</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Number of transactions each client runs. Default is 10.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-T</option> <replaceable>seconds</replaceable></term>
|
|
|
|
<term><option>--time=</option><replaceable>seconds</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Run the test for this many seconds, rather than a fixed number of
|
2012-05-09 19:39:53 +02:00
|
|
|
transactions per client. <option>-t</option> and
|
|
|
|
<option>-T</option> are mutually exclusive.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term><option>-v</option></term>
|
2013-06-27 14:52:13 +02:00
|
|
|
<term><option>--vacuum-all</option></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Vacuum all four standard tables before running the test.
|
2017-10-09 03:44:17 +02:00
|
|
|
With neither <option>-n</option> nor <option>-v</option>, <application>pgbench</application> will vacuum the
|
|
|
|
<structname>pgbench_tellers</structname> and <structname>pgbench_branches</structname>
|
|
|
|
tables, and will truncate <structname>pgbench_history</structname>.
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2013-05-14 03:37:01 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>--aggregate-interval=<replaceable>seconds</replaceable></option></term>
|
2013-05-14 03:37:01 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-01-02 18:26:03 +01:00
|
|
|
Length of aggregation interval (in seconds). May be used only
|
|
|
|
with <option>-l</option> option. With this option, the log contains
|
|
|
|
per-interval summary data, as described below.
|
2013-05-14 03:37:01 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2017-06-08 18:12:31 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>--log-prefix=<replaceable>prefix</replaceable></option></term>
|
2017-06-08 18:12:31 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Set the filename prefix for the log files created by
|
2017-10-09 03:44:17 +02:00
|
|
|
<option>--log</option>. The default is <literal>pgbench_log</literal>.
|
2017-06-08 18:12:31 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2013-05-14 03:37:01 +02:00
|
|
|
<varlistentry>
|
2016-07-07 03:09:26 +02:00
|
|
|
<term><option>--progress-timestamp</option></term>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
When showing progress (option <option>-P</option>), use a timestamp
|
2016-07-07 03:09:26 +02:00
|
|
|
(Unix epoch) instead of the number of seconds since the
|
|
|
|
beginning of the run. The unit is in seconds, with millisecond
|
|
|
|
precision after the dot.
|
|
|
|
This helps compare logs generated by various tools.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>--sampling-rate=<replaceable>rate</replaceable></option></term>
|
2013-05-14 03:37:01 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Sampling rate, used when writing data into the log, to reduce the
|
|
|
|
amount of log generated. If this option is given, only the specified
|
|
|
|
fraction of transactions are logged. 1.0 means all transactions will
|
|
|
|
be logged, 0.05 means only 5% of the transactions will be logged.
|
|
|
|
</para>
|
|
|
|
<para>
|
|
|
|
Remember to take the sampling rate into account when processing the
|
|
|
|
log file. For example, when computing tps values, you need to multiply
|
|
|
|
the numbers accordingly (e.g. with 0.01 sample rate, you'll only get
|
|
|
|
1/100 of the actual tps).
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2010-05-25 17:55:28 +02:00
|
|
|
</variablelist>
|
|
|
|
</para>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
2010-05-25 17:55:28 +02:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2 id="pgbench-common-options">
|
|
|
|
<title>Common Options</title>
|
2010-05-25 17:55:28 +02:00
|
|
|
|
|
|
|
<para>
|
|
|
|
<application>pgbench</application> accepts the following command-line
|
|
|
|
common arguments:
|
|
|
|
|
|
|
|
<variablelist>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-h</option> <replaceable>hostname</replaceable></term>
|
|
|
|
<term><option>--host=</option><replaceable>hostname</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2010-08-17 06:37:21 +02:00
|
|
|
The database server's host name
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-p</option> <replaceable>port</replaceable></term>
|
|
|
|
<term><option>--port=</option><replaceable>port</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
The database server's port number
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-U</option> <replaceable>login</replaceable></term>
|
|
|
|
<term><option>--username=</option><replaceable>login</replaceable></term>
|
2010-05-25 17:55:28 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
2010-08-17 06:37:21 +02:00
|
|
|
The user name to connect as
|
2010-05-25 17:55:28 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
2012-06-18 01:44:00 +02:00
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-V</option></term>
|
|
|
|
<term><option>--version</option></term>
|
2012-06-18 01:44:00 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Print the <application>pgbench</application> version and exit.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
2017-10-09 03:44:17 +02:00
|
|
|
<term><option>-?</option></term>
|
|
|
|
<term><option>--help</option></term>
|
2012-06-18 01:44:00 +02:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Show help about <application>pgbench</application> command line
|
|
|
|
arguments, and exit.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
2010-05-25 17:55:28 +02:00
|
|
|
</variablelist>
|
|
|
|
</para>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
|
|
|
</refsect1>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect1>
|
|
|
|
<title>Notes</title>
|
|
|
|
|
|
|
|
<refsect2>
|
2017-10-09 03:44:17 +02:00
|
|
|
<title>What is the <quote>Transaction</quote> Actually Performed in <application>pgbench</application>?</title>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
<application>pgbench</application> executes test scripts chosen randomly
|
pgbench: Allow changing weights for scripts
Previously, all scripts had the same probability of being chosen when
multiple of them were specified via -b, -f, -N, -S. With this commit,
-b and -f now search for an "@" in the script name and use the integer
found after it as the drawing probability for that script.
(One disadvantage is that if you have script whose names contain @, you
are now forced to specify "@1" at the end; otherwise the name's @ is
confused with a weight separator. We don't expect many pgbench script
with @ in their names in the wild, so this shouldn't be too serious a
problem.)
While at it, rework the interface between addScript, process_file,
process_builtin, and findBuiltin. It had gotten a bit out of hand with
recent commits.
Author: Fabien Coelho
Reviewed-By: Andres Freund, Robert Haas, Álvaro Herrera, Michaël Paquier
Discussion: http://www.postgresql.org/message-id/alpine.DEB.2.10.1603160721240.1666@sto
2016-03-19 16:32:42 +01:00
|
|
|
from a specified list.
|
2017-10-09 03:44:17 +02:00
|
|
|
They include built-in scripts with <option>-b</option> and
|
|
|
|
user-provided custom scripts with <option>-f</option>.
|
pgbench: Allow changing weights for scripts
Previously, all scripts had the same probability of being chosen when
multiple of them were specified via -b, -f, -N, -S. With this commit,
-b and -f now search for an "@" in the script name and use the integer
found after it as the drawing probability for that script.
(One disadvantage is that if you have script whose names contain @, you
are now forced to specify "@1" at the end; otherwise the name's @ is
confused with a weight separator. We don't expect many pgbench script
with @ in their names in the wild, so this shouldn't be too serious a
problem.)
While at it, rework the interface between addScript, process_file,
process_builtin, and findBuiltin. It had gotten a bit out of hand with
recent commits.
Author: Fabien Coelho
Reviewed-By: Andres Freund, Robert Haas, Álvaro Herrera, Michaël Paquier
Discussion: http://www.postgresql.org/message-id/alpine.DEB.2.10.1603160721240.1666@sto
2016-03-19 16:32:42 +01:00
|
|
|
Each script may be given a relative weight specified after a
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>@</literal> so as to change its drawing probability.
|
|
|
|
The default weight is <literal>1</literal>.
|
|
|
|
Scripts with a weight of <literal>0</literal> are ignored.
|
2016-01-27 02:54:22 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
The default built-in transaction script (also invoked with <option>-b tpcb-like</option>)
|
|
|
|
issues seven commands per transaction over randomly chosen <literal>aid</literal>,
|
|
|
|
<literal>tid</literal>, <literal>bid</literal> and <literal>balance</literal>.
|
2016-01-27 02:54:22 +01:00
|
|
|
The scenario is inspired by the TPC-B benchmark, but is not actually TPC-B,
|
|
|
|
hence the name.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<orderedlist>
|
|
|
|
<listitem><para><literal>BEGIN;</literal></para></listitem>
|
|
|
|
<listitem><para><literal>UPDATE pgbench_accounts SET abalance = abalance + :delta WHERE aid = :aid;</literal></para></listitem>
|
|
|
|
<listitem><para><literal>SELECT abalance FROM pgbench_accounts WHERE aid = :aid;</literal></para></listitem>
|
|
|
|
<listitem><para><literal>UPDATE pgbench_tellers SET tbalance = tbalance + :delta WHERE tid = :tid;</literal></para></listitem>
|
|
|
|
<listitem><para><literal>UPDATE pgbench_branches SET bbalance = bbalance + :delta WHERE bid = :bid;</literal></para></listitem>
|
|
|
|
<listitem><para><literal>INSERT INTO pgbench_history (tid, bid, aid, delta, mtime) VALUES (:tid, :bid, :aid, :delta, CURRENT_TIMESTAMP);</literal></para></listitem>
|
|
|
|
<listitem><para><literal>END;</literal></para></listitem>
|
|
|
|
</orderedlist>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
If you select the <literal>simple-update</literal> built-in (also <option>-N</option>),
|
2016-01-27 02:54:22 +01:00
|
|
|
steps 4 and 5 aren't included in the transaction.
|
|
|
|
This will avoid update contention on these tables, but
|
|
|
|
it makes the test case even less like TPC-B.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
If you select the <literal>select-only</literal> built-in (also <option>-S</option>),
|
|
|
|
only the <command>SELECT</command> is issued.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2>
|
2010-03-23 05:09:17 +01:00
|
|
|
<title>Custom Scripts</title>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
<application>pgbench</application> has support for running custom
|
|
|
|
benchmark scenarios by replacing the default transaction script
|
|
|
|
(described above) with a transaction script read from a file
|
2017-10-09 03:44:17 +02:00
|
|
|
(<option>-f</option> option). In this case a <quote>transaction</quote>
|
2016-01-27 02:54:22 +01:00
|
|
|
counts as one execution of a script file.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
A script file contains one or more SQL commands terminated by
|
|
|
|
semicolons. Empty lines and lines beginning with
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>--</literal> are ignored. Script files can also contain
|
|
|
|
<quote>meta commands</quote>, which are interpreted by <application>pgbench</application>
|
2010-03-23 05:09:17 +01:00
|
|
|
itself, as described below.
|
|
|
|
</para>
|
|
|
|
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
<note>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Before <productname>PostgreSQL</productname> 9.6, SQL commands in script files
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
were terminated by newlines, and so they could not be continued across
|
2017-10-09 03:44:17 +02:00
|
|
|
lines. Now a semicolon is <emphasis>required</emphasis> to separate consecutive
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
SQL commands (though a SQL command does not need one if it is followed
|
|
|
|
by a meta command). If you need to create a script file that works with
|
2017-10-09 03:44:17 +02:00
|
|
|
both old and new versions of <application>pgbench</application>, be sure to write
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
each SQL command on a single line ending with a semicolon.
|
|
|
|
</para>
|
|
|
|
</note>
|
|
|
|
|
2010-03-23 05:09:17 +01:00
|
|
|
<para>
|
|
|
|
There is a simple variable-substitution facility for script files.
|
2017-09-04 19:45:20 +02:00
|
|
|
Variable names must consist of letters (including non-Latin letters),
|
|
|
|
digits, and underscores.
|
2017-10-09 03:44:17 +02:00
|
|
|
Variables can be set by the command-line <option>-D</option> option,
|
2010-03-23 05:09:17 +01:00
|
|
|
explained above, or by the meta commands explained below.
|
2017-10-09 03:44:17 +02:00
|
|
|
In addition to any variables preset by <option>-D</option> command-line options,
|
2013-06-14 22:31:44 +02:00
|
|
|
there are a few variables that are preset automatically, listed in
|
2017-11-23 15:39:47 +01:00
|
|
|
<xref linkend="pgbench-automatic-variables"/>. A value specified for these
|
2017-10-09 03:44:17 +02:00
|
|
|
variables using <option>-D</option> takes precedence over the automatic presets.
|
2010-03-23 05:09:17 +01:00
|
|
|
Once set, a variable's
|
|
|
|
value can be inserted into a SQL command by writing
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>:</literal><replaceable>variablename</replaceable>. When running more than
|
2010-03-23 05:09:17 +01:00
|
|
|
one client session, each session has its own set of variables.
|
|
|
|
</para>
|
|
|
|
|
2013-06-14 22:31:44 +02:00
|
|
|
<table id="pgbench-automatic-variables">
|
2016-06-07 16:38:48 +02:00
|
|
|
<title>Automatic Variables</title>
|
2013-06-14 22:31:44 +02:00
|
|
|
<tgroup cols="2">
|
|
|
|
<thead>
|
|
|
|
<row>
|
|
|
|
<entry>Variable</entry>
|
|
|
|
<entry>Description</entry>
|
|
|
|
</row>
|
|
|
|
</thead>
|
|
|
|
|
|
|
|
<tbody>
|
|
|
|
<row>
|
|
|
|
<entry> <literal>scale</literal> </entry>
|
|
|
|
<entry>current scale factor</entry>
|
|
|
|
</row>
|
|
|
|
|
|
|
|
<row>
|
|
|
|
<entry> <literal>client_id</literal> </entry>
|
|
|
|
<entry>unique number identifying the client session (starts from zero)</entry>
|
|
|
|
</row>
|
|
|
|
</tbody>
|
|
|
|
</tgroup>
|
|
|
|
</table>
|
|
|
|
|
2010-03-23 05:09:17 +01:00
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Script file meta commands begin with a backslash (<literal>\</literal>) and
|
2017-01-20 17:10:02 +01:00
|
|
|
normally extend to the end of the line, although they can be continued
|
|
|
|
to additional lines by writing backslash-return.
|
2010-03-23 05:09:17 +01:00
|
|
|
Arguments to a meta command are separated by white space.
|
|
|
|
These meta commands are supported:
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<variablelist>
|
2016-03-01 19:04:09 +01:00
|
|
|
<varlistentry id='pgbench-metacommand-set'>
|
2010-03-23 05:09:17 +01:00
|
|
|
<term>
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>\set <replaceable>varname</replaceable> <replaceable>expression</replaceable></literal>
|
2010-03-23 05:09:17 +01:00
|
|
|
</term>
|
|
|
|
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Sets variable <replaceable>varname</replaceable> to a value calculated
|
|
|
|
from <replaceable>expression</replaceable>.
|
|
|
|
The expression may contain integer constants such as <literal>5432</literal>,
|
|
|
|
double constants such as <literal>3.14159</literal>,
|
|
|
|
references to variables <literal>:</literal><replaceable>variablename</replaceable>,
|
|
|
|
unary operators (<literal>+</literal>, <literal>-</literal>) and binary operators
|
|
|
|
(<literal>+</literal>, <literal>-</literal>, <literal>*</literal>, <literal>/</literal>,
|
|
|
|
<literal>%</literal>) with their usual precedence and associativity,
|
|
|
|
<link linkend="pgbench-builtin-functions">function calls</link>, and
|
2016-03-01 19:04:09 +01:00
|
|
|
parentheses.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2015-03-02 20:21:41 +01:00
|
|
|
Examples:
|
2010-07-29 21:34:41 +02:00
|
|
|
<programlisting>
|
2010-03-23 05:09:17 +01:00
|
|
|
\set ntellers 10 * :scale
|
2017-01-20 17:10:02 +01:00
|
|
|
\set aid (1021 * random(1, 100000 * :scale)) % \
|
|
|
|
(100000 * :scale) + 1
|
2012-05-15 21:55:13 +02:00
|
|
|
</programlisting></para>
|
2010-03-23 05:09:17 +01:00
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>\sleep <replaceable>number</replaceable> [ us | ms | s ]</literal>
|
2010-03-23 05:09:17 +01:00
|
|
|
</term>
|
|
|
|
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
Causes script execution to sleep for the specified duration in
|
2017-10-09 03:44:17 +02:00
|
|
|
microseconds (<literal>us</literal>), milliseconds (<literal>ms</literal>) or seconds
|
|
|
|
(<literal>s</literal>). If the unit is omitted then seconds are the default.
|
|
|
|
<replaceable>number</replaceable> can be either an integer constant or a
|
|
|
|
<literal>:</literal><replaceable>variablename</replaceable> reference to a variable
|
2010-03-23 05:09:17 +01:00
|
|
|
having an integer value.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Example:
|
2010-07-29 21:34:41 +02:00
|
|
|
<programlisting>
|
2010-03-23 05:09:17 +01:00
|
|
|
\sleep 10 ms
|
2012-05-15 21:55:13 +02:00
|
|
|
</programlisting></para>
|
2010-03-23 05:09:17 +01:00
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>\setshell <replaceable>varname</replaceable> <replaceable>command</replaceable> [ <replaceable>argument</replaceable> ... ]</literal>
|
2010-03-23 05:09:17 +01:00
|
|
|
</term>
|
|
|
|
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Sets variable <replaceable>varname</replaceable> to the result of the shell command
|
|
|
|
<replaceable>command</replaceable> with the given <replaceable>argument</replaceable>(s).
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
The command must return an integer value through its standard output.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>command</replaceable> and each <replaceable>argument</replaceable> can be either
|
|
|
|
a text constant or a <literal>:</literal><replaceable>variablename</replaceable> reference
|
|
|
|
to a variable. If you want to use an <replaceable>argument</replaceable> starting
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
with a colon, write an additional colon at the beginning of
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>argument</replaceable>.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Example:
|
2010-07-29 21:34:41 +02:00
|
|
|
<programlisting>
|
2010-03-23 05:09:17 +01:00
|
|
|
\setshell variable_to_be_assigned command literal_argument :variable ::literal_starting_with_colon
|
2012-05-15 21:55:13 +02:00
|
|
|
</programlisting></para>
|
2010-03-23 05:09:17 +01:00
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
|
|
|
|
<varlistentry>
|
|
|
|
<term>
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>\shell <replaceable>command</replaceable> [ <replaceable>argument</replaceable> ... ]</literal>
|
2010-03-23 05:09:17 +01:00
|
|
|
</term>
|
|
|
|
|
|
|
|
<listitem>
|
|
|
|
<para>
|
SQL commands in pgbench scripts are now ended by semicolons, not newlines.
To allow multiline SQL commands in scripts, adopt the same rules psql uses
to decide what is the end of a SQL command, to wit, an unquoted semicolon
not encased in parentheses. Do this by importing the same flex lexer that
psql uses, since coping with stuff like dollar-quoted literals is hard to
get right without going the full nine yards.
This makes use of the infrastructure added in commit 0ea9efbe9ec1bf07 to
support independently-written flex lexers scanning the same PsqlScanState
input-buffer data structure. Since that infrastructure isn't very
friendly to ad-hoc parsing code such as strtok(), improve exprscan.l
so that it can parse either whitespace-separated words or expression
tokens, on demand, and rewrite pgbench.c's backslash-command parsing
code to always use the lexer to fetch tokens.
It's still the case that pgbench backslash commands extend to the end
of the line, no more and no less. That could be changed in a fairly
localized way now, and there was some interest in doing so, but it
seems like material for a separate patch.
In passing, make some marginal cleanups in syntax error reporting,
const-ify a few data structures that could use it, and run some of
this code through pgindent.
I can't tell whether the MSVC build scripts need to be taught explicitly
about the changes here or not, but the buildfarm will soon tell us.
Kyotaro Horiguchi and Tom Lane
2016-03-20 17:58:44 +01:00
|
|
|
Same as <literal>\setshell</literal>, but the result of the command
|
|
|
|
is discarded.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Example:
|
2010-07-29 21:34:41 +02:00
|
|
|
<programlisting>
|
2010-03-23 05:09:17 +01:00
|
|
|
\shell command literal_argument :variable ::literal_starting_with_colon
|
2012-05-15 21:55:13 +02:00
|
|
|
</programlisting></para>
|
2010-03-23 05:09:17 +01:00
|
|
|
</listitem>
|
|
|
|
</varlistentry>
|
|
|
|
</variablelist>
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2016-03-01 19:04:09 +01:00
|
|
|
<refsect2 id="pgbench-builtin-functions">
|
|
|
|
<title>Built-In Functions</title>
|
|
|
|
|
2016-06-07 16:38:48 +02:00
|
|
|
<para>
|
2017-11-23 15:39:47 +01:00
|
|
|
The functions listed in <xref linkend="pgbench-functions"/> are built
|
2017-10-09 03:44:17 +02:00
|
|
|
into <application>pgbench</application> and may be used in expressions appearing in
|
2016-06-07 16:38:48 +02:00
|
|
|
<link linkend="pgbench-metacommand-set"><literal>\set</literal></link>.
|
|
|
|
</para>
|
2016-03-01 19:04:09 +01:00
|
|
|
|
|
|
|
<!-- list pgbench functions in alphabetical order -->
|
2016-06-07 16:38:48 +02:00
|
|
|
<table id="pgbench-functions">
|
2016-03-01 19:04:09 +01:00
|
|
|
<title>pgbench Functions</title>
|
|
|
|
<tgroup cols="5">
|
|
|
|
<thead>
|
|
|
|
<row>
|
|
|
|
<entry>Function</entry>
|
|
|
|
<entry>Return Type</entry>
|
|
|
|
<entry>Description</entry>
|
|
|
|
<entry>Example</entry>
|
|
|
|
<entry>Result</entry>
|
|
|
|
</row>
|
|
|
|
</thead>
|
|
|
|
<tbody>
|
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>abs(<replaceable>a</replaceable>)</function></literal></entry>
|
|
|
|
<entry>same as <replaceable>a</replaceable></entry>
|
|
|
|
<entry>absolute value</entry>
|
|
|
|
<entry><literal>abs(-17)</literal></entry>
|
|
|
|
<entry><literal>17</literal></entry>
|
2016-03-01 19:04:09 +01:00
|
|
|
</row>
|
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>debug(<replaceable>a</replaceable>)</function></literal></entry>
|
|
|
|
<entry>same as <replaceable>a</replaceable> </entry>
|
|
|
|
<entry>print <replaceable>a</replaceable> to <systemitem>stderr</systemitem>,
|
|
|
|
and return <replaceable>a</replaceable></entry>
|
|
|
|
<entry><literal>debug(5432.1)</literal></entry>
|
|
|
|
<entry><literal>5432.1</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>double(<replaceable>i</replaceable>)</function></literal></entry>
|
|
|
|
<entry>double</entry>
|
|
|
|
<entry>cast to double</entry>
|
|
|
|
<entry><literal>double(5432)</literal></entry>
|
|
|
|
<entry><literal>5432.0</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
2016-03-01 19:04:09 +01:00
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>greatest(<replaceable>a</replaceable> [, <replaceable>...</replaceable> ] )</function></literal></entry>
|
|
|
|
<entry>double if any <replaceable>a</replaceable> is double, else integer</entry>
|
|
|
|
<entry>largest value among arguments</entry>
|
|
|
|
<entry><literal>greatest(5, 4, 3, 2)</literal></entry>
|
|
|
|
<entry><literal>5</literal></entry>
|
2016-03-01 19:04:09 +01:00
|
|
|
</row>
|
2016-05-06 23:48:56 +02:00
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>int(<replaceable>x</replaceable>)</function></literal></entry>
|
|
|
|
<entry>integer</entry>
|
|
|
|
<entry>cast to int</entry>
|
|
|
|
<entry><literal>int(5.4 + 3.8)</literal></entry>
|
|
|
|
<entry><literal>9</literal></entry>
|
2016-05-06 23:48:56 +02:00
|
|
|
</row>
|
2016-03-01 19:04:09 +01:00
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>least(<replaceable>a</replaceable> [, <replaceable>...</replaceable> ] )</function></literal></entry>
|
|
|
|
<entry>double if any <replaceable>a</replaceable> is double, else integer</entry>
|
|
|
|
<entry>smallest value among arguments</entry>
|
|
|
|
<entry><literal>least(5, 4, 3, 2.1)</literal></entry>
|
|
|
|
<entry><literal>2.1</literal></entry>
|
2016-03-01 19:04:09 +01:00
|
|
|
</row>
|
2016-03-29 02:45:57 +02:00
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>pi()</function></literal></entry>
|
|
|
|
<entry>double</entry>
|
|
|
|
<entry>value of the constant PI</entry>
|
|
|
|
<entry><literal>pi()</literal></entry>
|
|
|
|
<entry><literal>3.14159265358979323846</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>random(<replaceable>lb</replaceable>, <replaceable>ub</replaceable>)</function></literal></entry>
|
|
|
|
<entry>integer</entry>
|
|
|
|
<entry>uniformly-distributed random integer in <literal>[lb, ub]</literal></entry>
|
|
|
|
<entry><literal>random(1, 10)</literal></entry>
|
|
|
|
<entry>an integer between <literal>1</literal> and <literal>10</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>random_exponential(<replaceable>lb</replaceable>, <replaceable>ub</replaceable>, <replaceable>parameter</replaceable>)</function></literal></entry>
|
|
|
|
<entry>integer</entry>
|
|
|
|
<entry>exponentially-distributed random integer in <literal>[lb, ub]</literal>,
|
|
|
|
see below</entry>
|
|
|
|
<entry><literal>random_exponential(1, 10, 3.0)</literal></entry>
|
|
|
|
<entry>an integer between <literal>1</literal> and <literal>10</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>random_gaussian(<replaceable>lb</replaceable>, <replaceable>ub</replaceable>, <replaceable>parameter</replaceable>)</function></literal></entry>
|
|
|
|
<entry>integer</entry>
|
|
|
|
<entry>Gaussian-distributed random integer in <literal>[lb, ub]</literal>,
|
|
|
|
see below</entry>
|
|
|
|
<entry><literal>random_gaussian(1, 10, 2.5)</literal></entry>
|
|
|
|
<entry>an integer between <literal>1</literal> and <literal>10</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
2017-12-14 12:30:22 +01:00
|
|
|
<row>
|
|
|
|
<entry><literal><function>random_zipfian(<replaceable>lb</replaceable>, <replaceable>ub</replaceable>, <replaceable>parameter</replaceable>)</function></literal></entry>
|
|
|
|
<entry>integer</entry>
|
|
|
|
<entry>Zipfian-distributed random integer in <literal>[lb, ub]</literal>,
|
|
|
|
see below</entry>
|
|
|
|
<entry><literal>random_zipfian(1, 10, 1.5)</literal></entry>
|
|
|
|
<entry>an integer between <literal>1</literal> and <literal>10</literal></entry>
|
|
|
|
</row>
|
2016-03-29 02:45:57 +02:00
|
|
|
<row>
|
2017-10-09 03:44:17 +02:00
|
|
|
<entry><literal><function>sqrt(<replaceable>x</replaceable>)</function></literal></entry>
|
|
|
|
<entry>double</entry>
|
|
|
|
<entry>square root</entry>
|
|
|
|
<entry><literal>sqrt(2.0)</literal></entry>
|
|
|
|
<entry><literal>1.414213562</literal></entry>
|
2016-03-29 02:45:57 +02:00
|
|
|
</row>
|
2016-03-01 19:04:09 +01:00
|
|
|
</tbody>
|
|
|
|
</tgroup>
|
|
|
|
</table>
|
2016-03-29 02:45:57 +02:00
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
The <literal>random</literal> function generates values using a uniform
|
2016-03-29 02:45:57 +02:00
|
|
|
distribution, that is all the values are drawn within the specified
|
2017-10-09 03:44:17 +02:00
|
|
|
range with equal probability. The <literal>random_exponential</literal> and
|
|
|
|
<literal>random_gaussian</literal> functions require an additional double
|
2016-03-29 02:45:57 +02:00
|
|
|
parameter which determines the precise shape of the distribution.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<itemizedlist>
|
|
|
|
<listitem>
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
For an exponential distribution, <replaceable>parameter</replaceable>
|
2016-03-29 02:45:57 +02:00
|
|
|
controls the distribution by truncating a quickly-decreasing
|
2017-10-09 03:44:17 +02:00
|
|
|
exponential distribution at <replaceable>parameter</replaceable>, and then
|
2016-03-29 02:45:57 +02:00
|
|
|
projecting onto integers between the bounds.
|
|
|
|
To be precise, with
|
|
|
|
<literallayout>
|
|
|
|
f(x) = exp(-parameter * (x - min) / (max - min + 1)) / (1 - exp(-parameter))
|
|
|
|
</literallayout>
|
2017-10-09 03:44:17 +02:00
|
|
|
Then value <replaceable>i</replaceable> between <replaceable>min</replaceable> and
|
|
|
|
<replaceable>max</replaceable> inclusive is drawn with probability:
|
|
|
|
<literal>f(i) - f(i + 1)</literal>.
|
2016-03-29 02:45:57 +02:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
Intuitively, the larger the <replaceable>parameter</replaceable>, the more
|
|
|
|
frequently values close to <replaceable>min</replaceable> are accessed, and the
|
|
|
|
less frequently values close to <replaceable>max</replaceable> are accessed.
|
|
|
|
The closer to 0 <replaceable>parameter</replaceable> is, the flatter (more
|
2016-03-29 02:45:57 +02:00
|
|
|
uniform) the access distribution.
|
|
|
|
A crude approximation of the distribution is that the most frequent 1%
|
2017-10-09 03:44:17 +02:00
|
|
|
values in the range, close to <replaceable>min</replaceable>, are drawn
|
|
|
|
<replaceable>parameter</replaceable>% of the time.
|
|
|
|
The <replaceable>parameter</replaceable> value must be strictly positive.
|
2016-03-29 02:45:57 +02:00
|
|
|
</para>
|
|
|
|
</listitem>
|
|
|
|
|
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
For a Gaussian distribution, the interval is mapped onto a standard
|
|
|
|
normal distribution (the classical bell-shaped Gaussian curve) truncated
|
2017-10-09 03:44:17 +02:00
|
|
|
at <literal>-parameter</literal> on the left and <literal>+parameter</literal>
|
2016-03-29 02:45:57 +02:00
|
|
|
on the right.
|
|
|
|
Values in the middle of the interval are more likely to be drawn.
|
2017-10-09 03:44:17 +02:00
|
|
|
To be precise, if <literal>PHI(x)</literal> is the cumulative distribution
|
|
|
|
function of the standard normal distribution, with mean <literal>mu</literal>
|
|
|
|
defined as <literal>(max + min) / 2.0</literal>, with
|
2016-03-29 02:45:57 +02:00
|
|
|
<literallayout>
|
2016-05-31 19:56:25 +02:00
|
|
|
f(x) = PHI(2.0 * parameter * (x - mu) / (max - min + 1)) /
|
|
|
|
(2.0 * PHI(parameter) - 1)
|
2016-03-29 02:45:57 +02:00
|
|
|
</literallayout>
|
2017-10-09 03:44:17 +02:00
|
|
|
then value <replaceable>i</replaceable> between <replaceable>min</replaceable> and
|
|
|
|
<replaceable>max</replaceable> inclusive is drawn with probability:
|
|
|
|
<literal>f(i + 0.5) - f(i - 0.5)</literal>.
|
|
|
|
Intuitively, the larger the <replaceable>parameter</replaceable>, the more
|
2016-03-29 02:45:57 +02:00
|
|
|
frequently values close to the middle of the interval are drawn, and the
|
2017-10-09 03:44:17 +02:00
|
|
|
less frequently values close to the <replaceable>min</replaceable> and
|
|
|
|
<replaceable>max</replaceable> bounds. About 67% of values are drawn from the
|
|
|
|
middle <literal>1.0 / parameter</literal>, that is a relative
|
|
|
|
<literal>0.5 / parameter</literal> around the mean, and 95% in the middle
|
|
|
|
<literal>2.0 / parameter</literal>, that is a relative
|
|
|
|
<literal>1.0 / parameter</literal> around the mean; for instance, if
|
|
|
|
<replaceable>parameter</replaceable> is 4.0, 67% of values are drawn from the
|
2016-03-29 02:45:57 +02:00
|
|
|
middle quarter (1.0 / 4.0) of the interval (i.e. from
|
2017-10-09 03:44:17 +02:00
|
|
|
<literal>3.0 / 8.0</literal> to <literal>5.0 / 8.0</literal>) and 95% from
|
|
|
|
the middle half (<literal>2.0 / 4.0</literal>) of the interval (second and third
|
|
|
|
quartiles). The minimum <replaceable>parameter</replaceable> is 2.0 for performance
|
2016-03-29 02:45:57 +02:00
|
|
|
of the Box-Muller transform.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
2017-12-14 12:30:22 +01:00
|
|
|
<listitem>
|
|
|
|
<para>
|
|
|
|
<literal>random_zipfian</literal> generates an approximated bounded zipfian
|
|
|
|
distribution. For <replaceable>parameter</replaceable> in (0, 1), an
|
|
|
|
approximated algorithm is taken from
|
|
|
|
"Quickly Generating Billion-Record Synthetic Databases",
|
|
|
|
Jim Gray et al, SIGMOD 1994. For <replaceable>parameter</replaceable>
|
|
|
|
in (1, 1000), a rejection method is used, based on
|
|
|
|
"Non-Uniform Random Variate Generation", Luc Devroye, p. 550-551,
|
|
|
|
Springer 1986. The distribution is not defined when the parameter's
|
|
|
|
value is 1.0. The drawing performance is poor for parameter values
|
|
|
|
close and above 1.0 and on a small range.
|
|
|
|
</para>
|
|
|
|
<para>
|
|
|
|
<replaceable>parameter</replaceable>
|
|
|
|
defines how skewed the distribution is. The larger the <replaceable>parameter</replaceable>, the more
|
|
|
|
frequently values to the beginning of the interval are drawn.
|
|
|
|
The closer to 0 <replaceable>parameter</replaceable> is,
|
|
|
|
the flatter (more uniform) the access distribution.
|
|
|
|
</para>
|
|
|
|
</listitem>
|
2016-03-29 02:45:57 +02:00
|
|
|
</itemizedlist>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
As an example, the full definition of the built-in TPC-B-like
|
|
|
|
transaction is:
|
|
|
|
|
|
|
|
<programlisting>
|
|
|
|
\set aid random(1, 100000 * :scale)
|
|
|
|
\set bid random(1, 1 * :scale)
|
|
|
|
\set tid random(1, 10 * :scale)
|
|
|
|
\set delta random(-5000, 5000)
|
|
|
|
BEGIN;
|
|
|
|
UPDATE pgbench_accounts SET abalance = abalance + :delta WHERE aid = :aid;
|
|
|
|
SELECT abalance FROM pgbench_accounts WHERE aid = :aid;
|
|
|
|
UPDATE pgbench_tellers SET tbalance = tbalance + :delta WHERE tid = :tid;
|
|
|
|
UPDATE pgbench_branches SET bbalance = bbalance + :delta WHERE bid = :bid;
|
|
|
|
INSERT INTO pgbench_history (tid, bid, aid, delta, mtime) VALUES (:tid, :bid, :aid, :delta, CURRENT_TIMESTAMP);
|
|
|
|
END;
|
|
|
|
</programlisting>
|
|
|
|
|
|
|
|
This script allows each iteration of the transaction to reference
|
|
|
|
different, randomly-chosen rows. (This example also shows why it's
|
|
|
|
important for each client session to have its own variables —
|
|
|
|
otherwise they'd not be independently touching different rows.)
|
|
|
|
</para>
|
|
|
|
|
2016-03-01 19:04:09 +01:00
|
|
|
</refsect2>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2>
|
2011-01-29 19:00:18 +01:00
|
|
|
<title>Per-Transaction Logging</title>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
With the <option>-l</option> option (but without
|
2017-01-02 18:26:03 +01:00
|
|
|
the <option>--aggregate-interval</option> option),
|
2017-10-09 03:44:17 +02:00
|
|
|
<application>pgbench</application> writes information about each transaction
|
2013-01-31 07:53:58 +01:00
|
|
|
to a log file. The log file will be named
|
2017-10-09 03:44:17 +02:00
|
|
|
<filename><replaceable>prefix</replaceable>.<replaceable>nnn</replaceable></filename>,
|
|
|
|
where <replaceable>prefix</replaceable> defaults to <literal>pgbench_log</literal>, and
|
|
|
|
<replaceable>nnn</replaceable> is the PID of the
|
2017-01-02 18:26:03 +01:00
|
|
|
<application>pgbench</application> process.
|
2017-10-09 03:44:17 +02:00
|
|
|
The prefix can be changed by using the <option>--log-prefix</option> option.
|
|
|
|
If the <option>-j</option> option is 2 or higher, so that there are multiple
|
2017-01-02 18:26:03 +01:00
|
|
|
worker threads, each will have its own log file. The first worker will
|
2016-11-09 22:26:32 +01:00
|
|
|
use the same name for its log file as in the standard single worker case.
|
2010-03-23 05:09:17 +01:00
|
|
|
The additional log files for the other workers will be named
|
2017-10-09 03:44:17 +02:00
|
|
|
<filename><replaceable>prefix</replaceable>.<replaceable>nnn</replaceable>.<replaceable>mmm</replaceable></filename>,
|
|
|
|
where <replaceable>mmm</replaceable> is a sequential number for each worker starting
|
2017-01-02 18:26:03 +01:00
|
|
|
with 1.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
The format of the log is:
|
|
|
|
|
2010-07-29 21:34:41 +02:00
|
|
|
<synopsis>
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>client_id</replaceable> <replaceable>transaction_no</replaceable> <replaceable>time</replaceable> <replaceable>script_no</replaceable> <replaceable>time_epoch</replaceable> <replaceable>time_us</replaceable> <optional> <replaceable>schedule_lag</replaceable> </optional>
|
2010-07-29 21:34:41 +02:00
|
|
|
</synopsis>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2017-01-02 18:26:03 +01:00
|
|
|
where
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>client_id</replaceable> indicates which client session ran the transaction,
|
|
|
|
<replaceable>transaction_no</replaceable> counts how many transactions have been
|
2017-01-02 18:26:03 +01:00
|
|
|
run by that session,
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>time</replaceable> is the total elapsed transaction time in microseconds,
|
|
|
|
<replaceable>script_no</replaceable> identifies which script file was used (useful when
|
|
|
|
multiple scripts were specified with <option>-f</option> or <option>-b</option>),
|
|
|
|
and <replaceable>time_epoch</replaceable>/<replaceable>time_us</replaceable> are a
|
2017-01-02 18:26:03 +01:00
|
|
|
Unix-epoch time stamp and an offset
|
2014-08-30 17:52:36 +02:00
|
|
|
in microseconds (suitable for creating an ISO 8601
|
2015-09-11 03:22:21 +02:00
|
|
|
time stamp with fractional seconds) showing when
|
2014-10-13 19:25:56 +02:00
|
|
|
the transaction completed.
|
2017-10-09 03:44:17 +02:00
|
|
|
The <replaceable>schedule_lag</replaceable> field is the difference between the
|
2014-10-13 19:25:56 +02:00
|
|
|
transaction's scheduled start time, and the time it actually started, in
|
2017-10-09 03:44:17 +02:00
|
|
|
microseconds. It is only present when the <option>--rate</option> option is used.
|
|
|
|
When both <option>--rate</option> and <option>--latency-limit</option> are used,
|
|
|
|
the <replaceable>time</replaceable> for a skipped transaction will be reported as
|
|
|
|
<literal>skipped</literal>.
|
2010-03-23 05:09:17 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-01-02 18:26:03 +01:00
|
|
|
Here is a snippet of a log file generated in a single-client run:
|
2010-07-29 21:34:41 +02:00
|
|
|
<screen>
|
2016-05-31 19:56:25 +02:00
|
|
|
0 199 2241 0 1175850568 995598
|
|
|
|
0 200 2465 0 1175850568 998079
|
|
|
|
0 201 2513 0 1175850569 608
|
|
|
|
0 202 2038 0 1175850569 2663
|
2014-10-13 19:25:56 +02:00
|
|
|
</screen>
|
|
|
|
|
2017-10-09 03:44:17 +02:00
|
|
|
Another example with <literal>--rate=100</literal>
|
|
|
|
and <literal>--latency-limit=5</literal> (note the additional
|
|
|
|
<replaceable>schedule_lag</replaceable> column):
|
2014-10-13 19:25:56 +02:00
|
|
|
<screen>
|
2016-05-31 19:56:25 +02:00
|
|
|
0 81 4621 0 1412881037 912698 3005
|
|
|
|
0 82 6173 0 1412881037 914578 4304
|
|
|
|
0 83 skipped 0 1412881037 914578 5217
|
|
|
|
0 83 skipped 0 1412881037 914578 5099
|
|
|
|
0 83 4722 0 1412881037 916203 3108
|
|
|
|
0 84 4142 0 1412881037 918023 2333
|
|
|
|
0 85 2465 0 1412881037 919759 740
|
2014-10-13 19:25:56 +02:00
|
|
|
</screen>
|
2016-04-23 20:48:02 +02:00
|
|
|
In this example, transaction 82 was late, because its latency (6.173 ms) was
|
2014-10-13 19:25:56 +02:00
|
|
|
over the 5 ms limit. The next two transactions were skipped, because they
|
|
|
|
were already late before they were even started.
|
|
|
|
</para>
|
2012-10-03 14:37:42 +02:00
|
|
|
|
|
|
|
<para>
|
|
|
|
When running a long test on hardware that can handle a lot of transactions,
|
2017-10-09 03:44:17 +02:00
|
|
|
the log files can become very large. The <option>--sampling-rate</option> option
|
2012-10-03 14:37:42 +02:00
|
|
|
can be used to log only a random sample of transactions.
|
|
|
|
</para>
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
2010-03-23 05:09:17 +01:00
|
|
|
|
2013-01-31 07:53:58 +01:00
|
|
|
<refsect2>
|
|
|
|
<title>Aggregated Logging</title>
|
2015-03-03 05:28:31 +01:00
|
|
|
|
2013-01-31 07:53:58 +01:00
|
|
|
<para>
|
2017-01-02 18:26:03 +01:00
|
|
|
With the <option>--aggregate-interval</option> option, a different
|
|
|
|
format is used for the log files:
|
2013-01-31 07:53:58 +01:00
|
|
|
|
|
|
|
<synopsis>
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>interval_start</replaceable> <replaceable>num_transactions</replaceable> <replaceable>sum_latency</replaceable> <replaceable>sum_latency_2</replaceable> <replaceable>min_latency</replaceable> <replaceable>max_latency</replaceable> <optional> <replaceable>sum_lag</replaceable> <replaceable>sum_lag_2</replaceable> <replaceable>min_lag</replaceable> <replaceable>max_lag</replaceable> <optional> <replaceable>skipped</replaceable> </optional> </optional>
|
2013-01-31 07:53:58 +01:00
|
|
|
</synopsis>
|
|
|
|
|
2017-01-02 18:26:03 +01:00
|
|
|
where
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>interval_start</replaceable> is the start of the interval (as a Unix
|
2017-01-02 18:26:03 +01:00
|
|
|
epoch time stamp),
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>num_transactions</replaceable> is the number of transactions
|
2017-01-02 18:26:03 +01:00
|
|
|
within the interval,
|
|
|
|
<replaceable>sum_latency</replaceable> is the sum of the transaction
|
|
|
|
latencies within the interval,
|
|
|
|
<replaceable>sum_latency_2</replaceable> is the sum of squares of the
|
|
|
|
transaction latencies within the interval,
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>min_latency</replaceable> is the minimum latency within the interval,
|
2017-01-02 18:26:03 +01:00
|
|
|
and
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>max_latency</replaceable> is the maximum latency within the interval.
|
2017-01-02 18:26:03 +01:00
|
|
|
The next fields,
|
2017-10-09 03:44:17 +02:00
|
|
|
<replaceable>sum_lag</replaceable>, <replaceable>sum_lag_2</replaceable>, <replaceable>min_lag</replaceable>,
|
|
|
|
and <replaceable>max_lag</replaceable>, are only present if the <option>--rate</option>
|
2017-01-02 18:26:03 +01:00
|
|
|
option is used.
|
|
|
|
They provide statistics about the time each transaction had to wait for the
|
2014-09-11 11:40:01 +02:00
|
|
|
previous one to finish, i.e. the difference between each transaction's
|
|
|
|
scheduled start time and the time it actually started.
|
2017-10-09 03:44:17 +02:00
|
|
|
The very last field, <replaceable>skipped</replaceable>,
|
|
|
|
is only present if the <option>--latency-limit</option> option is used, too.
|
2017-01-02 18:26:03 +01:00
|
|
|
It counts the number of transactions skipped because they would have
|
|
|
|
started too late.
|
|
|
|
Each transaction is counted in the interval when it was committed.
|
2013-01-31 07:53:58 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-01-02 18:26:03 +01:00
|
|
|
Here is some example output:
|
2013-01-31 07:53:58 +01:00
|
|
|
<screen>
|
|
|
|
1345828501 5601 1542744 483552416 61 2573
|
|
|
|
1345828503 7884 1979812 565806736 60 1479
|
|
|
|
1345828505 7208 1979422 567277552 59 1391
|
|
|
|
1345828507 7685 1980268 569784714 60 1398
|
|
|
|
1345828509 7073 1979779 573489941 236 1411
|
|
|
|
</screen></para>
|
|
|
|
|
|
|
|
<para>
|
2017-01-02 18:26:03 +01:00
|
|
|
Notice that while the plain (unaggregated) log file shows which script
|
|
|
|
was used for each transaction, the aggregated log does not. Therefore if
|
|
|
|
you need per-script data, you need to aggregate the data on your own.
|
2013-01-31 07:53:58 +01:00
|
|
|
</para>
|
|
|
|
|
|
|
|
</refsect2>
|
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2>
|
2011-01-29 19:00:18 +01:00
|
|
|
<title>Per-Statement Latencies</title>
|
2010-08-12 22:39:39 +02:00
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
With the <option>-r</option> option, <application>pgbench</application> collects
|
2010-08-12 22:39:39 +02:00
|
|
|
the elapsed transaction time of each statement executed by every
|
|
|
|
client. It then reports an average of those values, referred to
|
|
|
|
as the latency for each statement, after the benchmark has finished.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
For the default script, the output will look similar to this:
|
|
|
|
<screen>
|
|
|
|
starting vacuum...end.
|
2016-01-27 02:54:22 +01:00
|
|
|
transaction type: <builtin: TPC-B (sort of)>
|
2010-08-12 22:39:39 +02:00
|
|
|
scaling factor: 1
|
|
|
|
query mode: simple
|
|
|
|
number of clients: 10
|
|
|
|
number of threads: 1
|
|
|
|
number of transactions per client: 1000
|
|
|
|
number of transactions actually processed: 10000/10000
|
pgbench: Allow changing weights for scripts
Previously, all scripts had the same probability of being chosen when
multiple of them were specified via -b, -f, -N, -S. With this commit,
-b and -f now search for an "@" in the script name and use the integer
found after it as the drawing probability for that script.
(One disadvantage is that if you have script whose names contain @, you
are now forced to specify "@1" at the end; otherwise the name's @ is
confused with a weight separator. We don't expect many pgbench script
with @ in their names in the wild, so this shouldn't be too serious a
problem.)
While at it, rework the interface between addScript, process_file,
process_builtin, and findBuiltin. It had gotten a bit out of hand with
recent commits.
Author: Fabien Coelho
Reviewed-By: Andres Freund, Robert Haas, Álvaro Herrera, Michaël Paquier
Discussion: http://www.postgresql.org/message-id/alpine.DEB.2.10.1603160721240.1666@sto
2016-03-19 16:32:42 +01:00
|
|
|
latency average = 15.844 ms
|
|
|
|
latency stddev = 2.715 ms
|
2010-08-12 22:39:39 +02:00
|
|
|
tps = 618.764555 (including connections establishing)
|
|
|
|
tps = 622.977698 (excluding connections establishing)
|
pgbench: fix stats reporting when some transactions are skipped.
pgbench can skip some transactions when both -R and -L options are used.
Previously, this resulted in slightly silly statistics both in progress
reports and final output, because the skipped transactions were counted
as executed for TPS and related stats. Discount skipped xacts in TPS
numbers, and also when figuring the percentage of xacts exceeding the
latency limit.
Also, don't print per-script skipped-transaction counts when there is
only one script. That's redundant with the overall count, and it's
inconsistent with the fact that we don't print other per-script stats
when there's only one script. Clean up some unnecessary interactions
between what should be independent options that were due to that
decision.
While at it, avoid division-by-zero in cases where no transactions were
executed. While on modern platforms this would generally result in
printing "NaN" rather than a crash, that isn't spelled consistently
across platforms and it would confuse many people. Skip the relevant
output entirely when practical, else print zeroes.
Fabien Coelho, reviewed by Steve Singer, additional hacking by me
Discussion: https://postgr.es/m/26654.1505232433@sss.pgh.pa.us
2017-11-21 23:30:48 +01:00
|
|
|
statement latencies in milliseconds:
|
2016-03-29 19:13:51 +02:00
|
|
|
0.002 \set aid random(1, 100000 * :scale)
|
|
|
|
0.005 \set bid random(1, 1 * :scale)
|
|
|
|
0.002 \set tid random(1, 10 * :scale)
|
|
|
|
0.001 \set delta random(-5000, 5000)
|
|
|
|
0.326 BEGIN;
|
|
|
|
0.603 UPDATE pgbench_accounts SET abalance = abalance + :delta WHERE aid = :aid;
|
|
|
|
0.454 SELECT abalance FROM pgbench_accounts WHERE aid = :aid;
|
|
|
|
5.528 UPDATE pgbench_tellers SET tbalance = tbalance + :delta WHERE tid = :tid;
|
|
|
|
7.335 UPDATE pgbench_branches SET bbalance = bbalance + :delta WHERE bid = :bid;
|
|
|
|
0.371 INSERT INTO pgbench_history (tid, bid, aid, delta, mtime) VALUES (:tid, :bid, :aid, :delta, CURRENT_TIMESTAMP);
|
|
|
|
1.212 END;
|
2010-08-12 22:39:39 +02:00
|
|
|
</screen>
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
If multiple script files are specified, the averages are reported
|
|
|
|
separately for each script file.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
Note that collecting the additional timing information needed for
|
|
|
|
per-statement latency computation adds some overhead. This will slow
|
|
|
|
average execution speed and lower the computed TPS. The amount
|
|
|
|
of slowdown varies significantly depending on platform and hardware.
|
|
|
|
Comparing average TPS values with and without latency reporting enabled
|
|
|
|
is a good way to measure if the timing overhead is significant.
|
|
|
|
</para>
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
2010-08-12 22:39:39 +02:00
|
|
|
|
2012-05-09 19:39:53 +02:00
|
|
|
<refsect2>
|
2010-03-23 05:09:17 +01:00
|
|
|
<title>Good Practices</title>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
It is very easy to use <application>pgbench</application> to produce completely
|
2010-03-23 05:09:17 +01:00
|
|
|
meaningless numbers. Here are some guidelines to help you get useful
|
|
|
|
results.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
In the first place, <emphasis>never</emphasis> believe any test that runs
|
|
|
|
for only a few seconds. Use the <option>-t</option> or <option>-T</option> option
|
2010-03-23 05:09:17 +01:00
|
|
|
to make the run last at least a few minutes, so as to average out noise.
|
|
|
|
In some cases you could need hours to get numbers that are reproducible.
|
|
|
|
It's a good idea to try the test run a few times, to find out if your
|
|
|
|
numbers are reproducible or not.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
For the default TPC-B-like test scenario, the initialization scale factor
|
2017-10-09 03:44:17 +02:00
|
|
|
(<option>-s</option>) should be at least as large as the largest number of
|
|
|
|
clients you intend to test (<option>-c</option>); else you'll mostly be
|
|
|
|
measuring update contention. There are only <option>-s</option> rows in
|
|
|
|
the <structname>pgbench_branches</structname> table, and every transaction wants to
|
|
|
|
update one of them, so <option>-c</option> values in excess of <option>-s</option>
|
2010-03-23 05:09:17 +01:00
|
|
|
will undoubtedly result in lots of transactions blocked waiting for
|
|
|
|
other transactions.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
|
|
|
The default test scenario is also quite sensitive to how long it's been
|
|
|
|
since the tables were initialized: accumulation of dead rows and dead space
|
|
|
|
in the tables changes the results. To understand the results you must keep
|
|
|
|
track of the total number of updates and when vacuuming happens. If
|
|
|
|
autovacuum is enabled it can result in unpredictable changes in measured
|
|
|
|
performance.
|
|
|
|
</para>
|
|
|
|
|
|
|
|
<para>
|
2017-10-09 03:44:17 +02:00
|
|
|
A limitation of <application>pgbench</application> is that it can itself become
|
2010-03-23 05:09:17 +01:00
|
|
|
the bottleneck when trying to test a large number of client sessions.
|
2017-10-09 03:44:17 +02:00
|
|
|
This can be alleviated by running <application>pgbench</application> on a different
|
2010-03-23 05:09:17 +01:00
|
|
|
machine from the database server, although low network latency will be
|
2017-10-09 03:44:17 +02:00
|
|
|
essential. It might even be useful to run several <application>pgbench</application>
|
2010-03-23 05:09:17 +01:00
|
|
|
instances concurrently, on several client machines, against the same
|
|
|
|
database server.
|
|
|
|
</para>
|
2012-05-09 19:39:53 +02:00
|
|
|
</refsect2>
|
|
|
|
</refsect1>
|
|
|
|
</refentry>
|