PGBENCH(1)

📦 version 18.6, in opensuse-leap 16.0

okay bestie, sit down, hydrate, because we're doing benchmarks 💖 This is pgbench(1), straight out of the PostgreSQL 18.6 Documentation, and honestly? It's giving main character energy. Let's get into it.

🌸 NAME 🌸

pgbench — run a benchmark test on PostgreSQL

✨ SYNOPSIS ✨

pgbench -i [option...] [dbname]

pgbench [option...] [dbname]

💅 DESCRIPTION 💅

So pgbench is a simple little program for running benchmark tests on PostgreSQL. It runs the same sequence of SQL commands over and over, possibly in multiple concurrent database sessions, and then it calculates the average transaction rate (transactions per second). By default, pgbench tests a scenario that is loosely based on TPC-B, involving five SELECT, UPDATE, and INSERT commands per transaction. But fr fr, it is easy to test other cases by writing your own transaction script files. Slay. 💅

Typical output from pgbench looks like:

transaction type: <builtin: TPC-B (sort of)>
scaling factor: 10
query mode: simple
number of clients: 10
number of threads: 1
maximum number of tries: 1
number of transactions per client: 1000
number of transactions actually processed: 10000/10000
number of failed transactions: 0 (0.000%)
latency average = 11.013 ms
latency stddev = 7.351 ms
initial connection time = 45.758 ms
tps = 896.967014 (without initial connection time)

The first seven lines report some of the most important parameter settings. The sixth line reports the maximum number of tries for transactions with serialization or deadlock errors (see Failures and Serialization/Deadlock Retries for more information). The eighth 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 failed before completion or some SQL command(s) failed. (In -T mode, only the actual number of transactions is printed.) The next line reports the number of failed transactions due to serialization or deadlock errors (see Failures and Serialization/Deadlock Retries for more information). The last line reports the number of transactions per second.

The default TPC-B-like transaction test requires specific tables to be set up beforehand. pgbench should be invoked with the -i (initialize) option to create and populate these 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:

pgbench -i [ other-options ] dbname

where dbname is the name of the already-created database to test in. (You may also need -h, -p, and/or -U options to specify how to connect to the database server.)

🚨 Caution 🚨 pgbench -i creates four tables pgbench_accounts, pgbench_branches, pgbench_history, and pgbench_tellers, destroying any existing tables of these names. Be very careful to use another database if you have tables having these names! Not me watching you nuke your prod tables 💀

At the default “scale factor” of 1, the tables initially contain this many rows:

table                   # of rows
---------------------------------
pgbench_branches        1
pgbench_tellers         10
pgbench_accounts        100000
pgbench_history         0

You can (and, for most purposes, probably should) increase the number of rows by using the -s (scale factor) option. The -F (fillfactor) option might also be used at this point. ✨

Once you have done the necessary setup, you can run your benchmark with a command that doesn't include -i, that is

pgbench [ options ] dbname

In nearly all cases, you'll need some options to make a useful test. The most important options are -c (number of clients), -t (number of transactions), -T (time limit), and -f (specify a custom script file). See below for a full list. 👑

🎀 OPTIONS 🎀

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. Bestie, it's a whole vibe.

🦋 Initialization Options 🦋

pgbench accepts the following command-line initialization arguments:

[-d] dbname, [--dbname=]dbname — Specifies the name of the database to test in. If this is not specified, the environment variable PGDATABASE is used. If that is not set, the user name specified for the connection is used.
-i, --initialize — Required to invoke initialization mode. This is the one that makes the tables happen.
-I init_steps, --init-steps=init_steps — Perform just a selected set of the normal initialization steps. init_steps specifies the initialization steps to be performed, using one character per step. Each step is invoked in the specified order. The default is dtgvp. The available steps are:
- d (Drop) — Drop any existing pgbench tables.
- t (create Tables) — Create the tables used by the standard pgbench scenario, namely pgbench_accounts, pgbench_branches, pgbench_history, and pgbench_tellers.
- g or G (Generate data, client-side or server-side) — Generate data and load it into the standard tables, replacing any data already present.
  - With g (client-side data generation), data is generated in pgbench client and then sent to the server. This uses the client/server bandwidth extensively through a COPY. pgbench uses the FREEZE option to load data into ordinary (non-partition) tables with version 14 or later of PostgreSQL to speed up subsequent VACUUM. Using g causes logging to print one message every 100,000 rows while generating data for all tables.
  - With G (server-side data generation), only small queries are sent from the pgbench client and then data is actually generated in the server. No significant bandwidth is required for this variant, but the server will do more work. Using G causes logging not to print any progress message while generating data.
  - The default initialization behavior uses client-side data generation (equivalent to g).
- v (Vacuum) — Invoke VACUUM on the standard tables.
- p (create Primary keys) — Create primary key indexes on the standard tables.
- f (create Foreign keys) — Create foreign key constraints between the standard tables. (Note that this step is not performed by default.)
-F fillfactor, --fillfactor=fillfactor — Create the pgbench_accounts, pgbench_tellers and pgbench_branches tables with the given fillfactor. Default is 100.
-n, --no-vacuum — Perform no vacuuming during initialization. (This option suppresses the v initialization step, even if it was specified in -I.)
-q, --quiet — Switch logging to quiet mode, producing only one progress message per 5 seconds. The default logging prints one message each 100,000 rows, which often outputs many lines per second (especially on good hardware). No shade, but that's a lot of lines.

This setting has no effect if G is specified in -I.

-s scale_factor, --scale=scale_factor — Multiply the number of rows generated by the scale factor. For example, -s 100 will create 10,000,000 rows in the pgbench_accounts table. Default is 1. When the scale is 20,000 or larger, the columns used to hold account identifiers (aid columns) will switch to using larger integers (bigint), in order to be big enough to hold the range of account identifiers.
--foreign-keys — Create foreign key constraints between the standard tables. (This option adds the f step to the initialization step sequence, if it is not already present.)
--index-tablespace=index_tablespace — Create indexes in the specified tablespace, rather than the default tablespace.
--partition-method=NAME — Create a partitioned pgbench_accounts table with NAME method. Expected values are range or hash. This option requires that --partitions is set to non-zero. If unspecified, default is range.
--partitions=NUM — Create a partitioned pgbench_accounts table with NUM partitions of nearly equal size for the scaled number of accounts. Default is 0, meaning no partitioning.
--tablespace=tablespace — Create tables in the specified tablespace, rather than the default tablespace.
--unlogged-tables — Create all tables as unlogged tables, rather than permanent tables.

🌈 Benchmarking Options 🌈

pgbench accepts the following command-line benchmarking arguments:

-b scriptname[@weight], --builtin=scriptname[@weight] — Add the specified built-in script to the list of scripts to be executed. Available built-in scripts are: tpcb-like, simple-update and select-only. Unambiguous prefixes of built-in names are accepted. With the special name list, show the list of built-in scripts and exit immediately.

Optionally, write an integer weight after @ to adjust the probability of selecting this script versus other ones. The default weight is 1. See below for details.

-c clients, --client=clients — Number of clients simulated, that is, number of concurrent database sessions. Default is 1.
-C, --connect — Establish a new connection for each transaction, rather than doing it just once per client session. This is useful to measure the connection overhead.
-D varname=value, --define=varname=value — Define a variable for use by a custom script (see below). Multiple -D options are allowed.
-f filename[@weight], --file=filename[@weight] — Add a transaction script read from filename to the list of scripts to be executed.

Optionally, write an integer weight after @ to adjust the probability of selecting this script versus other ones. The default weight is 1. (To use a script file name that includes an @ character, append a weight so that there is no ambiguity, for example filen@me@1.) See below for details.

-j threads, --jobs=threads — Number of worker threads within pgbench. Using more than one thread can be helpful on multi-CPU machines. Clients are distributed as evenly as possible among available threads. Default is 1.
-l, --log — Write information about each transaction to a log file. See below for details.
-L limit, --latency-limit=limit — Transactions that last more than limit milliseconds are counted and reported separately, as late.

When throttling is used (--rate=...), transactions that lag behind schedule by more than limit ms, and thus have no hope of meeting the latency limit, are not sent to the server at all. They are counted and reported separately as skipped.

When the --max-tries option is in play, a transaction that dies from a serialization anomaly or a deadlock will not get retried if the total time of all its tries is greater than limit ms. And if you wanna cap only the time of the tries and not how many there are? Use --max-tries=0. Default behaviour: --max-tries is set to 1, so transactions with serialization/deadlock errors just don't get retried at all. Want the full tea on retrying these? See Failures and Serialization/Deadlock Retries 💅

-M querymode, --protocol=querymode — which protocol we use to ship queries to the server, bestie. Pick your fighter:
- simple — use the simple query protocol.
- extended — use the extended query protocol.
- prepared — use the extended query protocol with prepared statements.

In prepared mode, pgbench reuses the parse analysis result starting from the second query iteration, so it runs faster than the other modes. The default is the simple query protocol. (See Chapter 54 for more info, iykyk 📖)

-n, --no-vacuum — no vacuuming before the test runs, periodt. You need this option if you're running a custom test scenario that doesn't include the standard tables pgbench_accounts, pgbench_branches, pgbench_history, and pgbench_tellers.
-N, --skip-some-updates — run the built-in simple-update script. It's just shorthand for -b simple-update ✨
-P sec, --progress=sec — show a progress report every sec seconds. That report is stacked: the time since the beginning of the run, the TPS since the last report, and the transaction latency average, standard deviation, and number of failed transactions since the last report. Under throttling (-R), the latency is computed against the transaction's scheduled start time, not when it actually began — so it also folds in the average schedule lag time. And when --max-tries is used to enable retries after serialization/deadlock errors, the report also includes the number of retried transactions and the sum of all retries. So much data, we love a thorough queen 👑
-r, --report-per-command — after the benchmark finishes, report these stats for each command: the average per-statement latency (that's the execution time from the client's perspective), the number of failures, and the number of retries after serialization or deadlock errors in this command. Heads up: the retry stats only show up if --max-tries is not equal to 1. 🫶
-R rate, --rate=rate — execute transactions targeting the specified rate instead of going as fast as possible (which is the default). The rate is in transactions per second. If your target rate is above the maximum possible rate, the rate limit won't impact the results, no cap.

The rate is targeted by starting transactions along a Poisson-distributed schedule time line. The expected start time schedule moves forward based on when the client first started, not when the previous transaction ended. Which means: when transactions blow past their original scheduled end time, later ones can catch up again. Iconic.

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 one to finish. That wait time is called the schedule lag time, and its average and maximum get reported separately too. Want the latency with respect to the actual transaction start time — like, the time actually spent executing in the database? Just subtract the schedule lag time from the reported latency. Math is math 💅

If --latency-limit is used together with --rate, a transaction can lag behind so much that it's already over the latency limit the moment the previous transaction ends (because latency is calculated from the scheduled start time). Those transactions don't get sent to the server at all — they're skipped altogether and counted separately.

A high schedule lag time is your sign 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 falls further and further behind, and the schedule lag time keeps growing the longer the test runs. When that happens? You gotta reduce the specified transaction rate. It's giving "the system said no" 😭

-s scale_factor, --scale=scale_factor — report the specified scale factor in pgbench's output. With the built-in tests you don't need this — the correct scale factor gets detected by counting the number of rows in the pgbench_branches table. But if you're testing only custom benchmarks (the -f option), the scale factor will be reported as 1 unless you use this option. So use it, bestie ✨
-S, --select-only — run the built-in select-only script. Shorthand for -b select-only 🌸
-t transactions, --transactions=transactions — number of transactions each client runs. Default is 10.
-T seconds, --time=seconds — run the test for this many seconds instead of a fixed number of transactions per client. Heads up: -t and -T are mutually exclusive, pick one 🚫
-v, --vacuum-all — vacuum all four standard tables before running the test. And if you pass neither -n nor -v, pgbench will vacuum the pgbench_tellers and pgbench_branches tables, and will truncate pgbench_history. That's the middle-path vibe.
--aggregate-interval=seconds — length of the aggregation interval (in seconds). May be used only with the -l option. With this, the log contains per-interval summary data, as described below.
--exit-on-abort — bail immediately when any client is aborted due to some error. Without this option, even when a client gets aborted, other clients keep running as specified by -t or -T, and pgbench will print an incomplete result in that case. 🙈

Note: serialization failures or deadlock failures do not abort the client, so they are not affected by this option. See Failures and Serialization/Deadlock Retries for more info.

--failures-detailed — report failures in the per-transaction and aggregation logs, plus in the main and per-script reports, grouped by these types:
- serialization failures;
- deadlock failures;

See Failures and Serialization/Deadlock Retries for more. 🔥

--log-prefix=prefix — set the filename prefix for the log files created by --log. Default is pgbench_log.
--max-tries=number_of_tries — enable retries for transactions with serialization/deadlock errors and set the maximum number of these tries. This option can be combined with --latency-limit, which limits the total time of all transaction tries. Also important: you cannot use an unlimited number of tries (--max-tries=0) without --latency-limit or --time. Default value is 1, and transactions with serialization/deadlock errors are not retried. See Failures and Serialization/Deadlock Retries for the whole story 💀
--progress-timestamp — when showing progress (option -P), use a timestamp (Unix epoch) instead of the number of seconds since the beginning of the run. The unit is seconds, with millisecond precision after the dot. This helps compare logs generated by various tools. Iconic for the log girlies 🎀
--random-seed=seed — set the random generator seed. This seeds the system random number generator, which then produces a sequence of initial generator states, one for each thread. Values for seed can be:
- time — the default, the seed is based on the current time;
- rand — use a strong random source, failing if none is available;
- or an unsigned decimal integer value.

The random generator gets invoked explicitly from a pgbench script (random... functions) or implicitly (for instance, option --rate uses it to schedule transactions). When explicitly set, the value used for seeding is shown on the terminal. Any value allowed for seed may also be provided through the environment variable PGBENCH_RANDOM_SEED. To make sure the provided seed impacts all possible uses, put this option first or use the environment variable. ✨

Setting the seed explicitly lets you reproduce a pgbench run exactly, as far as random numbers are concerned. Since the random state is managed per thread, that means the exact same pgbench run for an identical invocation — if there's one client per thread and there are no external or data dependencies. From a statistical viewpoint, though, reproducing runs exactly is a bad idea, because it can hide performance variability or improve performance unduly, e.g. by hitting the same pages as a previous run. That said, it can also be a huge help for debugging — like re-running a tricky case that leads to an error. Use wisely. 🧚

--sampling-rate=rate — sampling rate, used when writing data into the log, to reduce the amount of log generated. Give this option and only the specified fraction of transactions gets logged. 1.0 means all transactions will be logged, 0.05 means only 5% of the transactions will be logged.

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). Don't get caught out by the math, bestie 📊

--show-script=scriptname — show the actual code of the builtin script scriptname on stderr, and exit immediately.
--verbose-errors — print messages about all errors and failures (errors without retrying), including which limit for retries was exceeded and how far it was exceeded for the serialization/deadlock failures.

Note: in this case the output can be significantly increased 😳 See Failures and Serialization/Deadlock Retries for more information.

💖 Common Options 💖

pgbench also accepts the following common command-line arguments for connection parameters and other common settings:

--debug — print debugging output.
-h hostname, --host=hostname — the database server's host name.
-p port, --port=port — the database server's port number.
-U login, --username=login — the user name to connect as.
-V, --version — print the pgbench version and exit.
-?, --help — show help about pgbench command line arguments, and exit. 🫶

💥 EXIT STATUS 💥

Ok so bestie, let's talk about how pgbench says goodbye, because it's giving very specific:

0 — the run went off without a hitch, she ate, no notes. Success.
1 — static problems. Like invalid command-line options, or internal errors which are, and I quote, supposed to never occur. Early errors that pop up when you're just starting the benchmark, like initial connection failures, also exit with status 1. Basically: she couldn't even get to the party.
2 — errors during the run. Database errors, problems in the script, that whole vibe. And in that case, pgbench will still print partial results if the --exit-on-abort option is not specified. So you get your half-eaten results, bestie. 🫶

🌿 ENVIRONMENT 🌿

PGDATABASE
PGHOST
PGPORT
PGUSER

Default connection parameters. The classics. ✨

This utility, like most other PostgreSQL utilities, uses the environment variables supported by libpq (see Section 32.15).

And then there's the glow-up variable: PG_COLOR specifies whether to use color in diagnostic messages. Possible values are always, auto and never. 💅

📝 NOTES 📝

🎀 What Is the “Transaction” Actually Performed in pgbench? 🎀

pgbench executes test scripts chosen randomly from a specified list. The scripts may include built-in scripts specified with -b and user-provided scripts specified with -f. Each script may be given a relative weight specified after an @ so as to change its selection probability. The default weight is 1. Scripts with a weight of 0 are ignored. (Zero weight = ghosted. 👻)

The default built-in transaction script (also invoked with -b tpcb-like) issues seven commands per transaction over randomly chosen aid, tid, bid and delta. The scenario is inspired by the TPC-B benchmark, but is not actually TPC-B, hence the name. It's giving TPC-B cosplay. 🙈

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;

If you select the simple-update built-in (also -N), 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. Like, even more not-TPC-B. 😭

If you select the select-only built-in (also -S), only the SELECT is issued. Minimalist. Main character energy. ✨

🦋 Custom Scripts 🦋

pgbench has support for running custom benchmark scenarios by replacing the default transaction script (described above) with a transaction script read from a file (-f option). In this case a “transaction” counts as one execution of a script file. One file run = one transaction, periodt.

A script file contains one or more SQL commands terminated by semicolons. Empty lines and lines beginning with -- are ignored. Script files can also contain “meta commands”, which are interpreted by pgbench itself, as described below.

Note 📌 Before PostgreSQL 9.6, SQL commands in script files were terminated by newlines, and so they could not be continued across lines. Now a semicolon is required to separate consecutive SQL commands (though an 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 both old and new versions of pgbench, be sure to write each SQL command on a single line ending with a semicolon. Nostalgia is a trap, bestie. 💀

It is assumed that pgbench scripts do not contain incomplete blocks of SQL transactions. If at runtime the client reaches the end of the script without completing the last transaction block, it will be aborted. ⚠️

There is a simple variable-substitution facility for script files. Variable names must consist of letters (including non-Latin letters), digits, and underscores, with the first character not being a digit. Variables can be set by the command-line -D option, explained above, or by the meta commands explained below. In addition to any variables preset by -D command-line options, there are a few variables that are preset automatically, listed in Table 301. A value specified for these variables using -D takes precedence over the automatic presets. Once set, a variable's value can be inserted into an SQL command by writing :variablename. When running more than one client session, each session has its own set of variables. pgbench supports up to 255 variable uses in one statement. (255, babe. Don't push it. 🙅)

Table 301. pgbench Automatic Variables ⭐

client_id — unique number identifying the client session (starts from zero)
default_seed — seed used in hash and pseudorandom permutation functions by default
random_seed — random generator seed (unless overwritten with -D)
scale — current scale factor

Script file meta commands begin with a backslash (\) and normally extend to the end of the line, although they can be continued to additional lines by writing backslash-return. Arguments to a meta command are separated by white space. These meta commands are supported:

\gset [prefix] and \aset [prefix]

These commands may be used to end SQL queries, taking the place of the terminating semicolon (;). Iconic substitution. 💅

When the \gset command is used, the preceding SQL query is expected to return one row, the columns of which are stored into variables named after column names, and prefixed with prefix if provided.

When the \aset command is used, all combined SQL queries (separated by \;) have their columns stored into variables named after column names, and prefixed with prefix if provided. If a query returns no row, no assignment is made and the variable can be tested for existence to detect this. If a query returns more than one row, the last value is kept.

\gset and \aset cannot be used in pipeline mode, since the query results are not yet available by the time the commands would need them. Timing is everything, and this timing? Doesn't work. 😭

The following example puts the final account balance from the first query into variable abalance, and fills variables p_two and p_three with integers from the third query. The result of the second query is discarded. The result of the two last combined queries are stored in variables four and five.

UPDATE pgbench_accounts
  SET abalance = abalance + :delta
  WHERE aid = :aid
  RETURNING abalance \gset
-- compound of two queries
SELECT 1 \;
SELECT 2 AS two, 3 AS three \gset p_
SELECT 4 AS four \; SELECT 5 AS five \aset

\if expression
\elif expression
\else
\endif

This group of commands implements nestable conditional blocks, similarly to psql's \if expression. Conditional expressions are identical to those with \set, with non-zero values interpreted as true. Choose-your-own-adventure but for SQL. 🧚

\set varname expression

Sets variable varname to a value calculated from expression. The expression may contain the NULL constant, Boolean constants TRUE and FALSE, integer constants such as 5432, double constants such as 3.14159, references to variables :variablename, operators with their usual SQL precedence and associativity, function calls, SQL CASE generic conditional expressions and parentheses.

Functions and most operators return NULL on NULL input.

For conditional purposes, non zero numerical values are TRUE, zero numerical values and NULL are FALSE.

Too large or small integer and double constants, as well as integer arithmetic operators (+, -, * and /) raise errors on overflows. Be careful, she will yell at you. 📣

When no final ELSE clause is provided to a CASE, the default value is NULL.

Examples:

\set ntellers 10 * :scale
\set aid (1021 * random(1, 100000 * :scale)) % \
           (100000 * :scale) + 1
\set divx CASE WHEN :x <> 0 THEN :y/:x ELSE NULL END

\sleep number [ us | ms | s ]

Causes script execution to sleep for the specified duration in microseconds (us), milliseconds (ms) or seconds (s). If the unit is omitted then seconds are the default. number can be either an integer constant or a :variablename reference to a variable having an integer value. Nap time. 😴

Example:

\sleep 10 ms

\setshell varname command [ argument ... ]

Sets variable varname to the result of the shell command command with the given argument(s). The command must return an integer value through its standard output.

command and each argument can be either a text constant or a :variablename reference to a variable. If you want to use an argument starting with a colon, write an additional colon at the beginning of argument. Double colon = escape hatch. 🕳️

Example:

\setshell variable_to_be_assigned command literal_argument :variable ::literal_starting_with_colon

\shell command [ argument ... ]

Same as \setshell, but the result of the command is discarded. We don't keep receipts here. 🙈

Example:

\shell command literal_argument :variable ::literal_starting_with_colon

\startpipeline
\syncpipeline
\endpipeline

This group of commands implements pipelining of SQL statements. A pipeline must begin with a \startpipeline and end with an \endpipeline. In between there may be any number of \syncpipeline commands, which sends a sync message without ending the ongoing pipeline and flushing the send buffer. In pipeline mode, statements are sent to the server without waiting for the results of previous statements. See Section 32.5 for more details. Pipeline mode requires the use of extended query protocol. 🔥

✨ Built-in Operators ✨

okay so pgbench is not out here making you import a whole math library, bestie — it's got a whole squad of operators built right in 💅 The arithmetic, bitwise, comparison and logical operators listed in Table 302 all come pre-loaded with pgbench, and you can drop them into any expression that shows up in \set. 🎀

vibe check on the ordering: the operators are listed in increasing precedence order, so the further down the list you go, the more precedence they're packing. iykyk 👑

and the type tea 🍓: except as noted, any operator that takes two numeric inputs will hand you back a double value if either of the inputs is a double. otherwise? you get an integer result. fr fr. 🌸

🌈 Table 302. pgbench Operators 🌈

okay so bestie, this is literally the whole menu of operators you can use in your pgbench scripts ✨ each entry shows you the types it takes in, the type it spits back out, what it actually does, and a lil example so it's not giving abstract math class energy 💅

boolean OR boolean → boolean — Logical OR. like, 5 or 0 → TRUE 🔥
boolean AND boolean → boolean — Logical AND. like, 3 and 0 → FALSE 💀
NOT boolean → boolean — Logical NOT. like, not false → TRUE ✨
boolean IS [NOT] (NULL|TRUE|FALSE) → boolean — Boolean value tests. like, 1 is null → FALSE 🫶
value ISNULL|NOTNULL → boolean — Nullness tests. like, 1 notnull → TRUE 👑
number = number → boolean — Equal. like, 5 = 4 → FALSE 😭
number <> number → boolean — Not equal. like, 5 <> 4 → TRUE 🌸
number != number → boolean — Not equal. like, 5 != 5 → FALSE 🙈
number < number → boolean — Less than. like, 5 < 4 → FALSE 💜
number <= number → boolean — Less than or equal to. like, 5 <= 4 → FALSE 🦋
number > number → boolean — Greater than. like, 5 > 4 → TRUE 🔥
number >= number → boolean — Greater than or equal to. like, 5 >= 4 → TRUE ⭐
integer | integer → integer — Bitwise OR. like, 1 | 2 → 3 🎀
integer # integer → integer — Bitwise XOR. like, 1 # 3 → 2 🌈
integer & integer → integer — Bitwise AND. like, 1 & 3 → 1 💅
~ integer → integer — Bitwise NOT. like, ~ 1 → -2 ✨
integer << integer → integer — Bitwise shift left. like, 1 << 2 → 4 🌸
integer >> integer → integer — Bitwise shift right. like, 8 >> 2 → 2 💖
number + number → number — Addition. like, 5 + 4 → 9 🔥
number - number → number — Subtraction. like, 3 - 2.0 → 1.0 👑
number * number → number — Multiplication. like, 5 * 4 → 20 💜
number / number → number — Division (truncates the result towards zero if both inputs are integers). like, 5 / 3 → 1 🫶
integer % integer → integer — Modulo (remainder). like, 3 % 2 → 1 🦋
- number → number — Negation. 🎀

       │               .PP - 2.0 → -2.0          │
       └─────────────────────────────────────────┘

🎀 Built-In Functions 🎀

okay so bestie, pgbench comes pre-loaded with its own little squad of functions 👑 — every single one of them is listed in Table 303, and they're built right in, no imports, no installs, nothing. ✨

You can drop them straight into any expression that shows up in \set. that's the vibe. 🫶

🌸 Table 303. pgbench Functions 🌸

okay so your pgbench script has a whole little toolbox of built-in functions and this is the entire menu, bestie — no skips, no gaps, every single one right here 💖

abs ( number ) → same type as input — absolute value, that's it, that's the vibe ✨
- e.g. abs(-17) → 17
debug ( number ) → same type as input — prints the argument to stderr, and then hands the argument right back to you 🫶
- e.g. debug(5432.1) → 5432.1
double ( number ) → double — casts to double 💅
- e.g. double(5432) → 5432.0
exp ( number ) → double — exponential, aka e raised to the given power 🌸
- e.g. exp(1.0) → 2.718281828459045
greatest ( number [, ... ] ) → double if any argument is double, else integer — selects the largest value among the arguments 👑
- e.g. greatest(5, 4, 3, 2) → 5
hash ( value [, seed ] ) → integer — this is literally just an alias for hash_murmur2, iykyk 🔥
- e.g. hash(10, 5432) → -5817877081768721676
hash_fnv1a ( value [, seed ] ) → integer — computes an FNV-1a hash ✨
- e.g. hash_fnv1a(10, 5432) → -7793829335365542153
hash_murmur2 ( value [, seed ] ) → integer — computes a MurmurHash2 hash 💜
- e.g. hash_murmur2(10, 5432) → -5817877081768721676
int ( number ) → integer — casts to integer 🎀
- e.g. int(5.4 + 3.8) → 9
least ( number [, ... ] ) → double if any argument is double, else integer — selects the smallest value among the arguments 🦋
- e.g. least(5, 4, 3, 2.1) → 2.1
ln ( number ) → double — natural logarithm 🌿
- e.g. ln(2.718281828459045) → 1.0
mod ( integer, integer ) → integer — modulo, aka the remainder 💫
- e.g. mod(54, 32) → 22
permute ( i, size [, seed ] ) → integer — the permuted value of i, in the range [0, size). This is the new position of i (modulo size) in a pseudorandom permutation of the integers 0...size-1, parameterized by seed, see below. 🌀
- e.g. permute(0, 4) → an integer between 0 and 3
pi () → double — approximate value of π 🥧
- e.g. pi() → 3.14159265358979323846
pow ( x, y ) → double / power ( x, y ) → double — x raised to the power of y, slay 🔥
- e.g. pow(2.0, 10) → 1024.0
random ( lb, ub ) → integer — computes a uniformly-distributed random integer in [lb, ub] 🎲
- e.g. random(1, 10) → an integer between 1 and 10
random_exponential ( lb, ub, parameter ) → integer — computes an exponentially-distributed random integer in [lb, ub], see below 📈
- e.g. random_exponential(1, 10, 3.0) → an integer between 1 and 10
random_gaussian ( lb, ub, parameter ) → integer — computes a Gaussian-distributed random integer in [lb, ub], see below 🔔
- e.g. random_gaussian(1, 10, 2.5) → an integer between 1 and 10
random_zipfian ( lb, ub, parameter ) → integer — spits out a Zipfian-distributed random integer in [lb, ub], deets below 👇

random_zipfian(1, 10, 1.5) → an integer between 1 and 10

sqrt ( number ) → double — square root, that's it, that's the whole job 🌱

sqrt(2.0) → 1.414213562

okay so the random function pulls values from a uniform distribution — meaning every value in the range gets the exact same chance, no favorites, all equal 💖. But random_exponential, random_gaussian and random_zipfian each want one extra double parameter, and that parameter is what decides the precise shape of the distribution. Bestie, the shape is everything.

Exponential 🍓 — here parameter shapes things by chopping off a fast-dropping exponential distribution right at parameter, then projecting that onto integers between the bounds. To be precise, with

f(x) = exp(-parameter * (x - min) / (max - min + 1)) / (1 - exp(-parameter))

then value i between min and max inclusive gets drawn with probability: f(i) - f(i + 1).

Vibe check: the bigger the parameter, the more often you hit values near min, and the less often you hit values near max. The closer parameter gets to 0, the flatter (aka more uniform) the access distribution becomes. Crude approximation of the distribution: the most frequent 1% of values in the range, close to min, get drawn parameter% of the time. And heads up — parameter must be strictly positive. No zero, no negatives, periodt 💅

Gaussian 🔔 — the interval gets mapped onto a standard normal distribution (the classic bell curve) truncated at -parameter on the left and +parameter on the right. Values in the middle of the interval are way more likely to be drawn. To be precise, if PHI(x) is the cumulative distribution function of the standard normal distribution, with mean mu defined as (max + min) / 2.0, with

f(x) = PHI(2.0 * parameter * (x - mu) / (max - min + 1)) /
       (2.0 * PHI(parameter) - 1)

then value i between min and max inclusive gets drawn with probability: f(i + 0.5) - f(i - 0.5). Intuitively: the bigger the parameter, the more frequently values close to the middle of the interval get drawn, and the less frequently values near min and max show up. About 67% of values come from the middle 1.0 / parameter, that's a relative 0.5 / parameter around the mean, and 95% from the middle 2.0 / parameter, that's a relative 1.0 / parameter around the mean. For instance, if parameter is 4.0, then 67% of values are drawn from the middle quarter (1.0 / 4.0) of the interval (i.e., from 3.0 / 8.0 to 5.0 / 8.0) and 95% from the middle half (2.0 / 4.0) of the interval (second and third quartiles). The minimum allowed parameter value is 2.0 — don't try to go lower, it's not the move 🙈

random_zipfian 👑 — generates a bounded Zipfian distribution. parameter defines how skewed the distribution is. The larger the parameter, the more frequently values closer to the beginning of the interval get drawn. The distribution is such that, assuming the range starts from 1, the ratio of the probability of drawing k versus drawing k+1 is ((k+1)/k)**parameter. For example, random_zipfian(1, ..., 2.5) produces the value 1 about (2/1)**2.5 = 5.66 times more frequently than 2, which itself is produced (3/2)**2.5 = 2.76 times more frequently than 3, and so on and so on, iconic behavior.

pgbench's implementation is based on "Non-Uniform Random Variate Generation", Luc Devroye, p. 550-551, Springer 1986. And because that algorithm has its limits, the parameter value is restricted to the range [1.001, 1000].

Note 🫶 When you're designing a benchmark that selects rows non-uniformly, be aware that the rows you choose may be correlated with other data — like IDs from a sequence, or the physical row ordering — and that can skew your performance measurements. To avoid this, you may wish to use the permute function, or some other additional step with a similar effect, to shuffle the selected rows and remove those correlations. Do not skip this, bestie, your numbers will lie to you 😭

The hash functions hash, hash_murmur2 and hash_fnv1a take an input value and an optional seed parameter. If you don't hand over a seed, the value of :default_seed gets used, which is initialized randomly unless you set it with the command-line -D option. ✨

permute accepts an input value, a size, and an optional seed parameter. It generates a pseudorandom permutation of integers in the range [0, size), and returns the index of the input value in the permuted values. The permutation chosen is parameterized by the seed, which defaults to :default_seed if you don't specify it. Unlike the hash functions, permute ensures that there are no collisions or holes in the output values — squeaky clean 🌈. Input values outside the interval are interpreted modulo the size. The function raises an error if the size is not positive. permute can be used to scatter the distribution of non-uniform random functions like random_zipfian or random_exponential so that values drawn more often aren't trivially correlated. For instance, the following pgbench script simulates a possible real-world workload typical for social media and blogging platforms, where a few accounts generate excessive load:

\set size 1000000
\set r random_zipfian(1, :size, 1.07)
\set k 1 + permute(:r, :size)

In some cases you need several distinct distributions that don't correlate with each other, and that's exactly when the optional seed parameter comes in handy:

\set k1 1 + permute(:r, :size, :default_seed + 123)
\set k2 1 + permute(:r, :size, :default_seed + 321)

A similar behavior can also be approximated with hash:

\set size 1000000
\set r random_zipfian(1, 100 * :size, 1.07)
\set k 1 + abs(hash(:r)) % :size

However — and this is the catch 🫠 — since hash generates collisions, some values will not be reachable at all and others will be more frequent than expected from the original distribution. So like, tradeoffs.

As an example, the full definition of the built-in TPC-B-like transaction is:

\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;

This script lets each iteration of the transaction 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. Main character energy per session, obviously 💅)

🌸 Per-Transaction Logging 🌸

With the -l option (but without the --aggregate-interval option), pgbench writes info about each transaction to a log file. The log file will be named prefix.nnn, where prefix defaults to pgbench_log, and nnn is the PID of the pgbench process. You can change the prefix with the --log-prefix option. If the -j option is 2 or higher, so that there are multiple worker threads, each one gets its own log file. The first worker will use the same name for its log file as in the standard single worker case. The additional log files for the other workers will be named prefix.nnn.mmm, where mmm is a sequential number for each worker starting with 1.

Each line in a log file describes one transaction. It contains the following space-separated fields:

client_id — identifies the client session that ran the transaction 🧚
transaction_no — counts how many transactions have been run by that session
time — transaction's elapsed time, in microseconds
script_no — identifies the script file that was used for the transaction (super useful when multiple scripts are specified with -f or -b)
time_epoch — transaction's completion time, as a Unix-epoch time stamp
time_us — fractional-second part of transaction's completion time, in microseconds
schedule_lag — transaction start delay, that is the difference between the transaction's scheduled start time and the time it actually started, in microseconds (present only if --rate is specified)
retries — count of retries after serialization or deadlock errors during the transaction (present only if --max-tries is not equal to one)

When both --rate and --latency-limit are used, the time for a skipped transaction will be reported as skipped. If the transaction ends with a failure, its time will be reported as failed. If you use the --failures-detailed option, the time of the failed transaction will be reported as serialization or deadlock depending on the type of failure (see Failures and Serialization/Deadlock Retries for more information).

Here's a snippet of a log file generated in a single-client run:

0 199 2241 0 1175850568 995598
0 200 2465 0 1175850568 998079
0 201 2513 0 1175850569 608
0 202 2038 0 1175850569 2663

Another example with --rate=100 and --latency-limit=5 (note the additional schedule_lag column):

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

In this example, transaction 82 was late, because its latency (6.173 ms) was over the 5 ms limit. The next two transactions were skipped, because they were already late before they were even started. Rough. 😭

okay so here's what a log file with failures and retries actually looks like, bestie — this one has the max number of tries set to 10, and note the extra retries column at the end (it's new, it's cute, it's informative):

3 0 47423 0 1499414498 34501 3
3 1 8333 0 1499414498 42848 0
3 2 8358 0 1499414498 51219 0
4 0 72345 0 1499414498 59433 6
1 3 41718 0 1499414498 67879 4
1 4 8416 0 1499414498 76311 0
3 3 33235 0 1499414498 84469 3
0 0 failed 0 1499414498 84905 9
2 0 failed 0 1499414498 86248 9
3 4 8307 0 1499414498 92788 0

And if you turn on the --failures-detailed option, the type of failure gets reported right there in the time column, like this:

3 0 47423 0 1499414498 34501 3
3 1 8333 0 1499414498 42848 0
3 2 8358 0 1499414498 51219 0
4 0 72345 0 1499414498 59433 6
1 3 41718 0 1499414498 67879 4
1 4 8416 0 1499414498 76311 0
3 3 33235 0 1499414498 84469 3
0 0 serialization 0 1499414498 84905 9
2 0 serialization 0 1499414498 86248 9
3 4 8307 0 1499414498 92788 0

real talk though: if you're running a long test on beefy hardware that can chew through a ton of transactions, those log files can get huge. That's where --sampling-rate comes in — it lets you log just a random sample of transactions instead of every single one. 💅

🌸 Aggregated Logging 🌸

When you pass --aggregate-interval, the log files use a whole different format. Now each log line describes one aggregation interval instead of one transaction. Every line is space-separated and contains these fields:

interval_start — start time of the interval, as a Unix-epoch time stamp
num_transactions — number of transactions within the interval
sum_latency — sum of transaction latencies
sum_latency_2 — sum of squares of transaction latencies
min_latency — minimum transaction latency
max_latency — maximum transaction latency
sum_lag — sum of transaction start delays (zero unless --rate is specified)
sum_lag_2 — sum of squares of transaction start delays (zero unless --rate is specified)
min_lag — minimum transaction start delay (zero unless --rate is specified)
max_lag — maximum transaction start delay (zero unless --rate is specified)
skipped — number of transactions skipped because they would have started too late (zero unless --rate and --latency-limit are specified)
retried — number of retried transactions (zero unless --max-tries is not equal to one)
retries — number of retries after serialization or deadlock errors (zero unless --max-tries is not equal to one)
serialization_failures — number of transactions that got a serialization error and were not retried afterwards (zero unless --failures-detailed is specified)
deadlock_failures — number of transactions that got a deadlock error and were not retried afterwards (zero unless --failures-detailed is specified)

Here's some example output generated with this option, look how pretty:

pgbench --aggregate-interval=10 --time=20 --client=10 --log --rate=1000 --latency-limit=10 --failures-detailed --max-tries=10 test

1650260552 5178 26171317 177284491527 1136 44462 2647617 7321113867 0 9866 64 7564 28340 4148 0
1650260562 4808 25573984 220121792172 1171 62083 3037380 9666800914 0 9998 598 7392 26621 4527 0

⚠️ heads up bestie — the plain (unaggregated) log format shows which script was used for each transaction, but the aggregated format does not. So if you need per-script data, you're gonna have to aggregate the data on your own. no cap. 😭

🎀 Per-Statement Report 🎀

With the -r option, pgbench collects these stats for each statement:

latency — elapsed transaction time for each statement. pgbench reports an average value of all successful runs of the statement.
The number of failures in this statement. See Failures and Serialization/Deadlock Retries for more information.
The number of retries after a serialization or a deadlock error in this statement. See Failures and Serialization/Deadlock Retries for more information.

quick note: the report only shows retry statistics if the --max-tries option is not equal to 1.

All values are computed for each statement executed by every client, and they get reported after the benchmark has finished. 🌸

For the default script, the output will look something like this:

starting vacuum...end.
transaction type: <builtin: TPC-B (sort of)>
scaling factor: 1
query mode: simple
number of clients: 10
number of threads: 1
maximum number of tries: 1
number of transactions per client: 1000
number of transactions actually processed: 10000/10000
number of failed transactions: 0 (0.000%)
number of transactions above the 50.0 ms latency limit: 1311/10000 (13.110 %)
latency average = 28.488 ms
latency stddev = 21.009 ms
initial connection time = 69.068 ms
tps = 346.224794 (without initial connection time)
statement latencies in milliseconds and failures:
   0.012  0  \set aid random(1, 100000 * :scale)
   0.002  0  \set bid random(1, 1 * :scale)
   0.002  0  \set tid random(1, 10 * :scale)
   0.002  0  \set delta random(-5000, 5000)
   0.319  0  BEGIN;
   0.834  0  UPDATE pgbench_accounts SET abalance = abalance + :delta WHERE aid = :aid;
   0.641  0  SELECT abalance FROM pgbench_accounts WHERE aid = :aid;
  11.126  0  UPDATE pgbench_tellers SET tbalance = tbalance + :delta WHERE tid = :tid;
  12.961  0  UPDATE pgbench_branches SET bbalance = bbalance + :delta WHERE bid = :bid;
   0.634  0  INSERT INTO pgbench_history (tid, bid, aid, delta, mtime) VALUES (:tid, :bid, :aid, :delta, CURRENT_TIMESTAMP);
   1.957  0  END;

And here's another example of output for the default script, this time using the serializable default transaction isolation level (PGOPTIONS='-c default_transaction_isolation=serializable' pgbench ...):

starting vacuum...end.
transaction type: <builtin: TPC-B (sort of)>
scaling factor: 1
query mode: simple
number of clients: 10
number of threads: 1
maximum number of tries: 10
number of transactions per client: 1000
number of transactions actually processed: 6317/10000
number of failed transactions: 3683 (36.830%)
number of transactions retried: 7667 (76.670%)
total number of retries: 45339
number of transactions above the 50.0 ms latency limit: 106/6317 (1.678 %)
latency average = 17.016 ms
latency stddev = 13.283 ms
initial connection time = 45.017 ms
tps = 186.792667 (without initial connection time)
statement latencies in milliseconds, failures and retries:
  0.006     0      0  \set aid random(1, 100000 * :scale)
  0.001     0      0  \set bid random(1, 1 * :scale)
  0.001     0      0  \set tid random(1, 10 * :scale)
  0.001     0      0  \set delta random(-5000, 5000)
  0.385     0      0  BEGIN;
  0.773     0      1  UPDATE pgbench_accounts SET abalance = abalance + :delta WHERE aid = :aid;
  0.624     0      0  SELECT abalance FROM pgbench_accounts WHERE aid = :aid;
  1.098   320   3762  UPDATE pgbench_tellers SET tbalance = tbalance + :delta WHERE tid = :tid;
  0.582  3363  41576  UPDATE pgbench_branches SET bbalance = bbalance + :delta WHERE bid = :bid;
  0.465     0      0  INSERT INTO pgbench_history (tid, bid, aid, delta, mtime) VALUES (:tid, :bid, :aid, :delta, CURRENT_TIMESTAMP);
  1.933     0      0  END;

If you specify multiple script files, all the statistics get reported separately for each script file. 🦋

⚠️ ngl bestie this matters — collecting the extra timing info needed for per-statement latency computation adds some overhead. It'll slow your average execution speed and lower the computed TPS. How much it slows down varies a lot depending on your platform and hardware. Comparing average TPS values with and without latency reporting enabled is a great way to measure whether the timing overhead is significant for you. 💅

💀 Failures and Serialization/Deadlock Retries 💀

When pgbench is executing, there are three main types of errors, and you gotta know which is which:

Errors of the main program. These are the most serious and always result in an immediate exit from pgbench with the corresponding error message. They include:
- errors at the beginning of pgbench (e.g. an invalid option value);
- errors in the initialization mode (e.g. the query to create tables for built-in scripts fails);
- errors before starting threads (e.g. could not connect to the database server, syntax error in the meta command, thread creation failure);
- internal pgbench errors (which are supposed to never occur...). 🙈
Errors when the thread manages its clients (e.g. the client could not start a connection to the database server / the socket for connecting the client to the database server has become invalid). In such cases all clients of this thread stop while the other threads keep working. However, if --exit-on-abort is specified, all of the threads stop immediately in this case.
Direct client errors. They lead to immediate exit from pgbench with the corresponding error message in the case of an internal pgbench error (which are supposed to never occur...) or when --exit-on-abort is specified. Otherwise, in the worst case, they only lead to the abortion of the failed client while the other clients continue their run (but some client errors are handled without an abortion of the client and reported separately, see below). Later in this section it's assumed that the errors we're discussing are only the direct client errors and they are not internal pgbench errors.

A client's run gets aborted the moment a serious error hits — like the connection to the database server dipped out, or the script ran all the way to the end without wrapping up the last transaction. Also, if an SQL or meta command fails for reasons that are not serialization or deadlock errors, the client is done-zo. But if an SQL command fails with serialization or deadlock errors? Bestie stays alive 💅. In those cases the current transaction is rolled back, which also means the client variables get set back to how they were before this transaction started running (we're assuming one transaction script holds just one transaction — peep What Is the "Transaction" Actually Performed in pgbench? if you want the full lore). Transactions with serialization or deadlock errors get repeated after rollbacks until they either complete successfully or hit the maximum number of tries (that's the --max-tries option) / the maximum time of retries (--latency-limit) / the end of the benchmark (--time). And if that very last trial run fails? The transaction gets reported as failed, but the client is not aborted — it just keeps on working, main character energy fr.

Note If you don't specify the --max-tries option, a transaction will never be retried after a serialization or deadlock error, because its default value is 1. So use an unlimited number of tries (--max-tries=0) plus the --latency-limit option to limit only the maximum time of tries. You can also use the --time option to cap the benchmark duration under an unlimited number of tries. 🫶 Be careful when repeating scripts that contain multiple transactions: the script is always retried completely, so successful transactions can end up being performed several times. 🙈 Be careful when repeating transactions with shell commands too. Unlike the results of SQL commands, the results of shell commands are not rolled back — except for the variable value of the \setshell command. 💀

The latency of a successful transaction includes the entire time of transaction execution with rollbacks and retries. But latency is measured only for successful transactions and commands, never for failed transactions or commands. Periodt.

The main report contains the number of failed transactions. And if the --max-tries option isn't equal to 1, the main report also contains statistics related to retries: the total number of retried transactions and the total number of retries. The per-script report inherits all these fields from the main report. The per-statement report displays retry statistics only if the --max-tries option isn't equal to 1.

Wanna group failures by basic types in per-transaction and aggregation logs, plus in the main and per-script reports? That's the --failures-detailed option, babe ✨. And if you also want to distinguish all errors and failures (errors without retrying) by type — including which limit for retries was exceeded and how much it was exceeded by, for the serialization/deadlock failures — reach for the --verbose-errors option.

✨ Table Access Methods ✨

So you can actually specify the Table Access Method for the pgbench tables 👑. The environment variable PGOPTIONS specifies database configuration options that get passed to PostgreSQL via the command line (See Section 19.1.4). Like, say you've got a hypothetical default Table Access Method for the tables that pgbench creates, called wuzza — you'd specify it like this:

PGOPTIONS='-c default_table_access_method=wuzza'

🌸 Good Practices 🌸

ngl it is very easy to use pgbench to produce completely meaningless numbers. So here are some guidelines to help you get results that actually mean something, bestie.

First things first: never believe any test that runs for only a few seconds. Use the -t or -T option to make the run last at least a few minutes, so the noise averages out. In some cases you could need hours to get numbers that are reproducible. And it's a good idea to try the test run a few times, to find out if your numbers are reproducible or not. 🦋

For the default TPC-B-like test scenario, the initialization scale factor (-s) should be at least as large as the largest number of clients you intend to test (-c) — otherwise you'll mostly be measuring update contention. There are only -s rows in the pgbench_branches table, and every transaction wants to update one of them, so -c values in excess of -s will undoubtedly result in lots of transactions blocked waiting for other transactions. 😭

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. And if autovacuum is enabled, it can result in unpredictable changes in measured performance. 🙈

A limitation of pgbench is that it can itself become the bottleneck when you're trying to test a large number of client sessions. You can alleviate that by running pgbench on a different machine from the database server — although low network latency will be essential. It might even be useful to run several pgbench instances concurrently, on several client machines, against the same database server. 💜

🔥 Security 🔥

If untrusted users have access to a database that hasn't adopted a secure schema usage pattern — do not run pgbench in that database. pgbench uses unqualified names and does not manipulate the search path. That's it, that's the whole warning. Stay safe out there 🫶

And that's the whole page, bestie!! This was PostgreSQL 18.6, dated 2026, and yeah — it's PGBENCH(1), iconic behavior only. Go forth and benchmark responsibly 💖✨🎀

📦 where this page comes from

opensuse-leap 16.0: postgresql18-contrib 18.6-160000.1.1 · PostgreSQL
opensuse-leap 16.0: postgresql18-contrib 18.6-160000.1.1 · PostgreSQL
opensuse-leap 16.0: postgresql18-contrib 18.6-160000.1.1 · PostgreSQL
opensuse-leap 16.0: postgresql18-contrib 18.6-160000.1.1 · PostgreSQL
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-10-13: a language model's rewrite of the original, under its terms