1POSTGRES(1) PostgreSQL 15.4 Documentation POSTGRES(1)
2
3
4
6 postgres - PostgreSQL database server
7
9 postgres [option...]
10
12 postgres is the PostgreSQL database server. In order for a client
13 application to access a database it connects (over a network or
14 locally) to a running postgres instance. The postgres instance then
15 starts a separate server process to handle the connection.
16
17 One postgres instance always manages the data of exactly one database
18 cluster. A database cluster is a collection of databases that is stored
19 at a common file system location (the “data area”). More than one
20 postgres instance can run on a system at one time, so long as they use
21 different data areas and different communication ports (see below).
22 When postgres starts it needs to know the location of the data area.
23 The location must be specified by the -D option or the PGDATA
24 environment variable; there is no default. Typically, -D or PGDATA
25 points directly to the data area directory created by initdb(1). Other
26 possible file layouts are discussed in Section 20.2.
27
28 By default postgres starts in the foreground and prints log messages to
29 the standard error stream. In practical applications postgres should be
30 started as a background process, perhaps at boot time.
31
32 The postgres command can also be called in single-user mode. The
33 primary use for this mode is during bootstrapping by initdb(1).
34 Sometimes it is used for debugging or disaster recovery; note that
35 running a single-user server is not truly suitable for debugging the
36 server, since no realistic interprocess communication and locking will
37 happen. When invoked in single-user mode from the shell, the user can
38 enter queries and the results will be printed to the screen, but in a
39 form that is more useful for developers than end users. In the
40 single-user mode, the session user will be set to the user with ID 1,
41 and implicit superuser powers are granted to this user. This user does
42 not actually have to exist, so the single-user mode can be used to
43 manually recover from certain kinds of accidental damage to the system
44 catalogs.
45
47 postgres accepts the following command-line arguments. For a detailed
48 discussion of the options consult Chapter 20. You can save typing most
49 of these options by setting up a configuration file. Some (safe)
50 options can also be set from the connecting client in an
51 application-dependent way to apply only for that session. For example,
52 if the environment variable PGOPTIONS is set, then libpq-based clients
53 will pass that string to the server, which will interpret it as
54 postgres command-line options.
55
56 General Purpose
57 -B nbuffers
58 Sets the number of shared buffers for use by the server processes.
59 The default value of this parameter is chosen automatically by
60 initdb. Specifying this option is equivalent to setting the
61 shared_buffers configuration parameter.
62
63 -c name=value
64 Sets a named run-time parameter. The configuration parameters
65 supported by PostgreSQL are described in Chapter 20. Most of the
66 other command line options are in fact short forms of such a
67 parameter assignment. -c can appear multiple times to set multiple
68 parameters.
69
70 -C name
71 Prints the value of the named run-time parameter, and exits. (See
72 the -c option above for details.) This returns values from
73 postgresql.conf, modified by any parameters supplied in this
74 invocation. It does not reflect parameters supplied when the
75 cluster was started.
76
77 This can be used on a running server for most parameters. However,
78 the server must be shut down for some runtime-computed parameters
79 (e.g., shared_memory_size, shared_memory_size_in_huge_pages, and
80 wal_segment_size).
81
82 This option is meant for other programs that interact with a server
83 instance, such as pg_ctl(1), to query configuration parameter
84 values. User-facing applications should instead use SHOW or the
85 pg_settings view.
86
87 -d debug-level
88 Sets the debug level. The higher this value is set, the more
89 debugging output is written to the server log. Values are from 1 to
90 5. It is also possible to pass -d 0 for a specific session, which
91 will prevent the server log level of the parent postgres process
92 from being propagated to this session.
93
94 -D datadir
95 Specifies the file system location of the database configuration
96 files. See Section 20.2 for details.
97
98 -e
99 Sets the default date style to “European”, that is DMY ordering of
100 input date fields. This also causes the day to be printed before
101 the month in certain date output formats. See Section 8.5 for more
102 information.
103
104 -F
105 Disables fsync calls for improved performance, at the risk of data
106 corruption in the event of a system crash. Specifying this option
107 is equivalent to disabling the fsync configuration parameter. Read
108 the detailed documentation before using this!
109
110 -h hostname
111 Specifies the IP host name or address on which postgres is to
112 listen for TCP/IP connections from client applications. The value
113 can also be a comma-separated list of addresses, or * to specify
114 listening on all available interfaces. An empty value specifies not
115 listening on any IP addresses, in which case only Unix-domain
116 sockets can be used to connect to the server. Defaults to listening
117 only on localhost. Specifying this option is equivalent to setting
118 the listen_addresses configuration parameter.
119
120 -i
121 Allows remote clients to connect via TCP/IP (Internet domain)
122 connections. Without this option, only local connections are
123 accepted. This option is equivalent to setting listen_addresses to
124 * in postgresql.conf or via -h.
125
126 This option is deprecated since it does not allow access to the
127 full functionality of listen_addresses. It's usually better to set
128 listen_addresses directly.
129
130 -k directory
131 Specifies the directory of the Unix-domain socket on which postgres
132 is to listen for connections from client applications. The value
133 can also be a comma-separated list of directories. An empty value
134 specifies not listening on any Unix-domain sockets, in which case
135 only TCP/IP sockets can be used to connect to the server. The
136 default value is normally /tmp, but that can be changed at build
137 time. Specifying this option is equivalent to setting the
138 unix_socket_directories configuration parameter.
139
140 -l
141 Enables secure connections using SSL. PostgreSQL must have been
142 compiled with support for SSL for this option to be available. For
143 more information on using SSL, refer to Section 19.9.
144
145 -N max-connections
146 Sets the maximum number of client connections that this server will
147 accept. The default value of this parameter is chosen automatically
148 by initdb. Specifying this option is equivalent to setting the
149 max_connections configuration parameter.
150
151 -p port
152 Specifies the TCP/IP port or local Unix domain socket file
153 extension on which postgres is to listen for connections from
154 client applications. Defaults to the value of the PGPORT
155 environment variable, or if PGPORT is not set, then defaults to the
156 value established during compilation (normally 5432). If you
157 specify a port other than the default port, then all client
158 applications must specify the same port using either command-line
159 options or PGPORT.
160
161 -s
162 Print time information and other statistics at the end of each
163 command. This is useful for benchmarking or for use in tuning the
164 number of buffers.
165
166 -S work-mem
167 Specifies the base amount of memory to be used by sorts and hash
168 tables before resorting to temporary disk files. See the
169 description of the work_mem configuration parameter in
170 Section 20.4.1.
171
172 -V
173 --version
174 Print the postgres version and exit.
175
176 --name=value
177 Sets a named run-time parameter; a shorter form of -c.
178
179 --describe-config
180 This option dumps out the server's internal configuration
181 variables, descriptions, and defaults in tab-delimited COPY format.
182 It is designed primarily for use by administration tools.
183
184 -?
185 --help
186 Show help about postgres command line arguments, and exit.
187
188 Semi-Internal Options
189 The options described here are used mainly for debugging purposes, and
190 in some cases to assist with recovery of severely damaged databases.
191 There should be no reason to use them in a production database setup.
192 They are listed here only for use by PostgreSQL system developers.
193 Furthermore, these options might change or be removed in a future
194 release without notice.
195
196 -f { s | i | o | b | t | n | m | h }
197 Forbids the use of particular scan and join methods: s and i
198 disable sequential and index scans respectively, o, b and t disable
199 index-only scans, bitmap index scans, and TID scans respectively,
200 while n, m, and h disable nested-loop, merge and hash joins
201 respectively.
202
203 Neither sequential scans nor nested-loop joins can be disabled
204 completely; the -fs and -fn options simply discourage the optimizer
205 from using those plan types if it has any other alternative.
206
207 -n
208 This option is for debugging problems that cause a server process
209 to die abnormally. The ordinary strategy in this situation is to
210 notify all other server processes that they must terminate and then
211 reinitialize the shared memory and semaphores. This is because an
212 errant server process could have corrupted some shared state before
213 terminating. This option specifies that postgres will not
214 reinitialize shared data structures. A knowledgeable system
215 programmer can then use a debugger to examine shared memory and
216 semaphore state.
217
218 -O
219 Allows the structure of system tables to be modified. This is used
220 by initdb.
221
222 -P
223 Ignore system indexes when reading system tables, but still update
224 the indexes when modifying the tables. This is useful when
225 recovering from damaged system indexes.
226
227 -t pa[rser] | pl[anner] | e[xecutor]
228 Print timing statistics for each query relating to each of the
229 major system modules. This option cannot be used together with the
230 -s option.
231
232 -T
233 This option is for debugging problems that cause a server process
234 to die abnormally. The ordinary strategy in this situation is to
235 notify all other server processes that they must terminate and then
236 reinitialize the shared memory and semaphores. This is because an
237 errant server process could have corrupted some shared state before
238 terminating. This option specifies that postgres will stop all
239 other server processes by sending the signal SIGSTOP, but will not
240 cause them to terminate. This permits system programmers to collect
241 core dumps from all server processes by hand.
242
243 -v protocol
244 Specifies the version number of the frontend/backend protocol to be
245 used for a particular session. This option is for internal use
246 only.
247
248 -W seconds
249 A delay of this many seconds occurs when a new server process is
250 started, after it conducts the authentication procedure. This is
251 intended to give an opportunity to attach to the server process
252 with a debugger.
253
254 Options for Single-User Mode
255 The following options only apply to the single-user mode (see Single-
256 User Mode below).
257
258 --single
259 Selects the single-user mode. This must be the first argument on
260 the command line.
261
262 database
263 Specifies the name of the database to be accessed. This must be the
264 last argument on the command line. If it is omitted it defaults to
265 the user name.
266
267 -E
268 Echo all commands to standard output before executing them.
269
270 -j
271 Use semicolon followed by two newlines, rather than just newline,
272 as the command entry terminator.
273
274 -r filename
275 Send all server log output to filename. This option is only honored
276 when supplied as a command-line option.
277
279 PGCLIENTENCODING
280 Default character encoding used by clients. (The clients can
281 override this individually.) This value can also be set in the
282 configuration file.
283
284 PGDATA
285 Default data directory location
286
287 PGDATESTYLE
288 Default value of the DateStyle run-time parameter. (The use of this
289 environment variable is deprecated.)
290
291 PGPORT
292 Default port number (preferably set in the configuration file)
293
295 A failure message mentioning semget or shmget probably indicates you
296 need to configure your kernel to provide adequate shared memory and
297 semaphores. For more discussion see Section 19.4. You might be able to
298 postpone reconfiguring your kernel by decreasing shared_buffers to
299 reduce the shared memory consumption of PostgreSQL, and/or by reducing
300 max_connections to reduce the semaphore consumption.
301
302 A failure message suggesting that another server is already running
303 should be checked carefully, for example by using the command
304
305 $ ps ax | grep postgres
306
307 or
308
309 $ ps -ef | grep postgres
310
311 depending on your system. If you are certain that no conflicting server
312 is running, you can remove the lock file mentioned in the message and
313 try again.
314
315 A failure message indicating inability to bind to a port might indicate
316 that that port is already in use by some non-PostgreSQL process. You
317 might also get this error if you terminate postgres and immediately
318 restart it using the same port; in this case, you must simply wait a
319 few seconds until the operating system closes the port before trying
320 again. Finally, you might get this error if you specify a port number
321 that your operating system considers to be reserved. For example, many
322 versions of Unix consider port numbers under 1024 to be “trusted” and
323 only permit the Unix superuser to access them.
324
326 The utility command pg_ctl(1) can be used to start and shut down the
327 postgres server safely and comfortably.
328
329 If at all possible, do not use SIGKILL to kill the main postgres
330 server. Doing so will prevent postgres from freeing the system
331 resources (e.g., shared memory and semaphores) that it holds before
332 terminating. This might cause problems for starting a fresh postgres
333 run.
334
335 To terminate the postgres server normally, the signals SIGTERM, SIGINT,
336 or SIGQUIT can be used. The first will wait for all clients to
337 terminate before quitting, the second will forcefully disconnect all
338 clients, and the third will quit immediately without proper shutdown,
339 resulting in a recovery run during restart.
340
341 The SIGHUP signal will reload the server configuration files. It is
342 also possible to send SIGHUP to an individual server process, but that
343 is usually not sensible.
344
345 To cancel a running query, send the SIGINT signal to the process
346 running that command. To terminate a backend process cleanly, send
347 SIGTERM to that process. See also pg_cancel_backend and
348 pg_terminate_backend in Section 9.27.2 for the SQL-callable equivalents
349 of these two actions.
350
351 The postgres server uses SIGQUIT to tell subordinate server processes
352 to terminate without normal cleanup. This signal should not be used by
353 users. It is also unwise to send SIGKILL to a server process — the main
354 postgres process will interpret this as a crash and will force all the
355 sibling processes to quit as part of its standard crash-recovery
356 procedure.
357
359 The -- options will not work on FreeBSD or OpenBSD. Use -c instead.
360 This is a bug in the affected operating systems; a future release of
361 PostgreSQL will provide a workaround if this is not fixed.
362
364 To start a single-user mode server, use a command like
365
366 postgres --single -D /usr/local/pgsql/data other-options my_database
367
368 Provide the correct path to the database directory with -D, or make
369 sure that the environment variable PGDATA is set. Also specify the name
370 of the particular database you want to work in.
371
372 Normally, the single-user mode server treats newline as the command
373 entry terminator; there is no intelligence about semicolons, as there
374 is in psql. To continue a command across multiple lines, you must type
375 backslash just before each newline except the last one. The backslash
376 and adjacent newline are both dropped from the input command. Note that
377 this will happen even when within a string literal or comment.
378
379 But if you use the -j command line switch, a single newline does not
380 terminate command entry; instead, the sequence
381 semicolon-newline-newline does. That is, type a semicolon immediately
382 followed by a completely empty line. Backslash-newline is not treated
383 specially in this mode. Again, there is no intelligence about such a
384 sequence appearing within a string literal or comment.
385
386 In either input mode, if you type a semicolon that is not just before
387 or part of a command entry terminator, it is considered a command
388 separator. When you do type a command entry terminator, the multiple
389 statements you've entered will be executed as a single transaction.
390
391 To quit the session, type EOF (Control+D, usually). If you've entered
392 any text since the last command entry terminator, then EOF will be
393 taken as a command entry terminator, and another EOF will be needed to
394 exit.
395
396 Note that the single-user mode server does not provide sophisticated
397 line-editing features (no command history, for example). Single-user
398 mode also does not do any background processing, such as automatic
399 checkpoints or replication.
400
402 To start postgres in the background using default values, type:
403
404 $ nohup postgres >logfile 2>&1 </dev/null &
405
406 To start postgres with a specific port, e.g., 1234:
407
408 $ postgres -p 1234
409
410 To connect to this server using psql, specify this port with the -p
411 option:
412
413 $ psql -p 1234
414
415 or set the environment variable PGPORT:
416
417 $ export PGPORT=1234
418 $ psql
419
420 Named run-time parameters can be set in either of these styles:
421
422 $ postgres -c work_mem=1234
423 $ postgres --work-mem=1234
424
425 Either form overrides whatever setting might exist for work_mem in
426 postgresql.conf. Notice that underscores in parameter names can be
427 written as either underscore or dash on the command line. Except for
428 short-term experiments, it's probably better practice to edit the
429 setting in postgresql.conf than to rely on a command-line switch to set
430 a parameter.
431
433 initdb(1), pg_ctl(1)
434
435
436
437PostgreSQL 15.4 2023 POSTGRES(1)