Initialization Options

pgbench accepts the following command-line initialization arguments:

Benchmarking Options

pgbench accepts the following command-line benchmarking arguments:

Common Options

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

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.

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.

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.

If you select the select-only built-in (also -S), only the SELECT is issued.

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.

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.

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 295. 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.

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:

Built-in Operators

The arithmetic, bitwise, comparison and logical operators listed in Table 296 are built into pgbench and may be used in expressions appearing in \set. The operators are listed in increasing precedence order. Except as noted, operators taking two numeric inputs will produce a double value if either input is double, otherwise they produce an integer result.

Built-In Functions

The functions listed in Table 297 are built into pgbench and may be used in expressions appearing in \set.

The random function generates values using a uniform distribution, that is all the values are drawn within the specified range with equal probability. The random_exponential, random_gaussian and random_zipfian functions require an additional double parameter which determines the precise shape of the distribution.

Hash functions hash, hash_murmur2 and hash_fnv1a accept an input value and an optional seed parameter. In case the seed isn't provided the value of :default_seed is used, which is initialized randomly unless set by 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 not specified. Unlike the hash functions, permute ensures that there are no collisions or holes in the output values. 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 such as random_zipfian or random_exponential so that values drawn more often are not 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 several distinct distributions are needed which don't correlate with each other and this is 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, since hash generates collisions, some values will not be reachable and others will be more frequent than expected from the original distribution.

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 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.)

Per-Transaction Logging

With the -l option (but without the --aggregate-interval option), pgbench writes information 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. The prefix can be changed by using the --log-prefix option. If the -j option is 2 or higher, so that there are multiple worker threads, each will have 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:

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 is 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.

The following example shows a snippet of a log file with failures and retries, with the maximum number of tries set to 10 (note the additional retries column):

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

If the --failures-detailed option is used, the type of failure is reported in the time 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

When running a long test on hardware that can handle a lot of transactions, the log files can become very large. The --sampling-rate option can be used to log only a random sample of transactions.

Aggregated Logging

With the --aggregate-interval option, a different format is used for the log files. Each log line describes one aggregation interval. It contains the following space-separated fields:

Here is some example output generated with these options:

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

Notice that while the plain (unaggregated) log format shows which script was used for each transaction, the aggregated format does not. Therefore if you need per-script data, you need to aggregate the data on your own.

Per-Statement Report

With the -r option, pgbench collects the following statistics for each statement:

The report displays retry statistics only if the --max-tries option is not equal to 1.

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

For the default script, the output will look similar to 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;

Another example of output for the default script using 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 multiple script files are specified, all statistics are reported separately for each script file.

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.

Failures and Serialization/Deadlock Retries

When executing pgbench, there are three main types of errors:

A client's run is aborted in case of a serious error; for example, the connection with the database server was lost or the end of script was reached without completing the last transaction. In addition, if execution of an SQL or meta command fails for reasons other than serialization or deadlock errors, the client is aborted. Otherwise, if an SQL command fails with serialization or deadlock errors, the client is not aborted. In such cases, the current transaction is rolled back, which also includes setting the client variables as they were before the run of this transaction (it is assumed that one transaction script contains only one transaction; see What Is the "Transaction" Actually Performed in pgbench? for more information). Transactions with serialization or deadlock errors are repeated after rollbacks until they complete successfully or reach the maximum number of tries (specified by the --max-tries option) / the maximum time of retries (specified by the --latency-limit option) / the end of benchmark (specified by the --time option). If the last trial run fails, this transaction will be reported as failed but the client is not aborted and continues to work.

The latency of a successful transaction includes the entire time of transaction execution with rollbacks and retries. The latency is measured only for successful transactions and commands but not for failed transactions or commands.

The main report contains the number of failed transactions. If the --max-tries option is not equal to 1, the main report also contains statistics related to retries: the total number of retried transactions and 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 is not equal to 1.

If you want to group failures by basic types in per-transaction and aggregation logs, as well as in the main and per-script reports, use the --failures-detailed option. 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, use the --verbose-errors option.

Good Practices

It is very easy to use pgbench to produce completely meaningless numbers. Here are some guidelines to help you get useful results.

In the first place, 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 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.

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); else 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. 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 trying to test a large number of client sessions. This can be alleviated 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 has not adopted a secure schema usage pattern, do not run pgbench in that database. pgbench uses unqualified names and does not manipulate the search path.