1POSTGRES(1) PostgreSQL Server Applications POSTGRES(1)
2
6 postgres - PostgreSQL database server
7
10 postgres [ option... ]
11
13 postgres is the PostgreSQL database server. In order for a client
14 application to access a database it connects (over a network or
15 locally) to a running postgres process. The postgres instance then
16 starts a separate server process to handle the connection.
17 One postgres instance always manages the data from exactly one database
19 cluster. A database cluster is a collection of databases that is stored
20 at a common file system location (the ``data area''). More than one
21 postgres process can run on a system at one time, so long as they use
22 different data areas and different communication ports (see below).
23 When postgres starts it needs to know the location of the data area.
24 The location must be specified by the -D option or the PGDATA environ‐
25 ment variable; there is no default. Typically, -D or PGDATA points
26 directly to the data area directory created by initdb(1). Other possi‐
27 ble file layouts are discussed in in the documentation.
28 By default postgres starts in the foreground and prints log messages to
30 the standard error stream. In practical applications postgres should be
31 started as a background process, perhaps at boot time.
32 The postgres command can also be called in single-user mode. The pri‐
34 mary use for this mode is during bootstrapping by initdb(1). Sometimes
35 it is used for debugging or disaster recovery (but note that running a
36 single-user server is not truly suitable for debugging the server,
37 since no realistic interprocess communication and locking will happen).
38 When invoked in single-user mode from the shell, the user can enter
39 queries and the results will be printed to the screen, but in a form
40 that is more useful for developers than end users. In the single-user
41 mode, the session user will be set to the user with ID 1, and implicit
42 superuser powers are granted to this user. This user does not actually
43 have to exist, so the single-user mode can be used to manually recover
44 from certain kinds of accidental damage to the system catalogs.
45
47 postgres accepts the following command-line arguments. For a detailed
48 discussion of the options consult in the documentation. You can save
49 typing most of these options by setting up a configuration file. Some
50 (safe) options can also be set from the connecting client in an appli‐
51 cation-dependent way to apply only for that session. For example, if
52 the environment variable PGOPTIONS is set, then libpq-based clients
53 will pass that string to the server, which will interpret it as post‐
54 gres command-line options.
55 GENERAL PURPOSE
57 -A 0|1 Enables run-time assertion checks, which is a debugging aid to
58 detect programming mistakes. This option is only available if
59 assertions were enabled when PostgreSQL was compiled. If so, the
60 default is on.
61 -B nbuffers
63 Sets the number of shared buffers for use by the server pro‐
64 cesses. The default value of this parameter is chosen automati‐
65 cally by initdb; refer to in the documentation for more informa‐
66 tion.
67 -c name=value
69 Sets a named run-time parameter. The configuration parameters
70 supported by PostgreSQL are described in in the documentation.
71 Most of the other command line options are in fact short forms
72 of such a parameter assignment. -c can appear multiple times to
73 set multiple parameters.
74 -d debug-level
76 Sets the debug level. The higher this value is set, the more
77 debugging output is written to the server log. Values are from 1
78 to 5. It is also possible to pass -d 0 for a specific session,
79 which will prevent the server log level of the parent postgres
80 process from being propagated to this session.
81 -D datadir
83 Specifies the file system location of the data directory or con‐
84 figuration file(s). See in the documentation for details.
85 -e Sets the default date style to ``European'', that is DMY order‐
87 ing of input date fields. This also causes the day to be printed
88 before the month in certain date output formats. See in the
89 documentation for more information.
90 -F Disables fsync calls for improved performance, at the risk of
92 data corruption in the event of a system crash. Specifying this
93 option is equivalent to disabling the fsync configuration param‐
94 eter. Read the detailed documentation before using this!
95 -h hostname
97 Specifies the IP host name or address on which postgres is to
98 listen for TCP/IP connections from client applications. The
99 value can also be a comma-separated list of addresses, or * to
100 specify listening on all available interfaces. An empty value
101 specifies not listening on any IP addresses, in which case only
102 Unix-domain sockets can be used to connect to the server.
103 Defaults to listening only on localhost. Specifying this option
104 is equivalent to setting the listen_addresses configuration
105 parameter.
106 -i Allows remote clients to connect via TCP/IP (Internet domain)
108 connections. Without this option, only local connections are
109 accepted. This option is equivalent to setting listen_addresses
110 to * in postgresql.conf or via -h.
111 This option is deprecated since it does not allow access to the
113 full functionality of listen_addresses. It's usually better to
114 set listen_addresses directly.
115 -k directory
117 Specifies the directory of the Unix-domain socket on which post‐
118 gres is to listen for connections from client applications. The
119 default is normally /tmp, but can be changed at build time.
120 -l Enables secure connections using SSL. PostgreSQL must have been
122 compiled with support for SSL for this option to be available.
123 For more information on using SSL, refer to in the documenta‐
124 tion.
125 -N max-connections
127 Sets the maximum number of client connections that this server
128 will accept. By default, this value is 32, but it can be set as
129 high as your system will support. (Note that -B is required to
130 be at least twice -N. See in the documentation for a discussion
131 of system resource requirements for large numbers of client con‐
132 nections.) Specifying this option is equivalent to setting the
133 max_connections configuration parameter.
134 -o extra-options
136 The command line-style options specified in extra-options are
137 passed to all server processes started by this postgres process.
138 If the option string contains any spaces, the entire string must
139 be quoted.
140 The use of this option is obsolete; all command-line options for
142 server processes can be specified directly on the postgres com‐
143 mand line.
144 -p port
146 Specifies the TCP/IP port or local Unix domain socket file
147 extension on which postgres is to listen for connections from
148 client applications. Defaults to the value of the PGPORT envi‐
149 ronment variable, or if PGPORT is not set, then defaults to the
150 value established during compilation (normally 5432). If you
151 specify a port other than the default port, then all client
152 applications must specify the same port using either command-
153 line options or PGPORT.
154 -s Print time information and other statistics at the end of each
156 command. This is useful for benchmarking or for use in tuning
157 the number of buffers.
158 -S work-mem
160 Specifies the amount of memory to be used by internal sorts and
161 hashes before resorting to temporary disk files. See the
162 description of the work_mem configuration parameter in in the
163 documentation.
164 --name=value
166 Sets a named run-time parameter; a shorter form of -c.
167 --describe-config
169 This option dumps out the server's internal configuration vari‐
170 ables, descriptions, and defaults in tab-delimited COPY format.
171 It is designed primarily for use by administration tools.
172 SEMI-INTERNAL OPTIONS
174 There are several other options that may be specified, used mainly for
175 debugging purposes and in some cases to assist with recovery of severe‐
176 ly damaged databases. There should be no reason to use them in a pro‐
177 duction database setup. These are listed here only for the use by Post‐
178 greSQL system developers. Furthermore, any of these options may disap‐
179 pear or change in a future release without notice.
180 -f { s | i | m | n | h }
182 Forbids the use of particular scan and join methods: s and i
183 disable sequential and index scans respectively, while n, m, and
184 h disable nested-loop, merge and hash joins respectively.
185 Neither sequential scans nor nested-loop joins can be disabled
187 completely; the -fs and -fn options simply discourage the opti‐
188 mizer from using those plan types if it has any other alterna‐
189 tive.
190 -n This option is for debugging problems that cause a server
192 process to die abnormally. The ordinary strategy in this situa‐
193 tion is to notify all other server processes that they must ter‐
194 minate and then reinitialize the shared memory and semaphores.
195 This is because an errant server process could have corrupted
196 some shared state before terminating. This option specifies that
197 postgres will not reinitialize shared data structures. A knowl‐
198 edgeable system programmer can then use a debugger to examine
199 shared memory and semaphore state.
200 -O Allows the structure of system tables to be modified. This is
202 used by initdb.
203 -P Ignore system indexes when reading system tables (but still
205 update the indexes when modifying the tables). This is useful
206 when recovering from damaged system indexes.
207 -t pa[rser] | pl[anner] | e[xecutor]
209 Print timing statistics for each query relating to each of the
210 major system modules. This option cannot be used together with
211 the -s option.
212 -T This option is for debugging problems that cause a server
214 process to die abnormally. The ordinary strategy in this situa‐
215 tion is to notify all other server processes that they must ter‐
216 minate and then reinitialize the shared memory and semaphores.
217 This is because an errant server process could have corrupted
218 some shared state before terminating. This option specifies that
219 postgres will stop all other server processes by sending the
220 signal SIGSTOP, but will not cause them to terminate. This per‐
221 mits system programmers to collect core dumps from all server
222 processes by hand.
223 -v protocol
225 Specifies the version number of the frontend/backend protocol to
226 be used for a particular session. This option is for internal
227 use only.
228 -W seconds
230 A delay of this many seconds occurs when a new server process is
231 started, after it conducts the authentication procedure. This
232 is intended to give an opportunity to attach to the server
233 process with a debugger.
234 -y database
236 Indicates that this is a subprocess started by a parent postgres
237 process, and specifies the database to use. This option is for
238 internal use only.
239 OPTIONS FOR SINGLE-USER MODE
241 The following options only apply to the single-user mode.
242 --single
244 Selects the single-user mode. This must be the first argument on
245 the command line.
246 database
248 Specifies the name of the database to be accessed. If it is
249 omitted it defaults to the user name.
250 -E Echo all commands.
252 -j Disables use of newline as a statement delimiter.
254 -r filename
256 Send all server log output to filename. In normal multiuser
257 mode, this option is ignored, and stderr is used by all pro‐
258 cesses.
259
261 PGCLIENTENCODING
262 Default character encoding used by clients. (The clients may
263 override this individually.) This value can also be set in the
264 configuration file.
265 PGDATA Default data directory location
267 PGDATESTYLE
269 Default value of the datestyle run-time parameter. (The use of
270 this environment variable is deprecated.)
271 PGPORT Default port (preferably set in the configuration file)
273 TZ Server time zone
275
277 A failure message mentioning semget or shmget probably indicates you
278 need to configure your kernel to provide adequate shared memory and
279 semaphores. For more discussion see in the documentation. You may be
280 able to postpone reconfiguring your kernel by decreasing shared_buffers
281 to reduce the shared memory consumption of PostgreSQL, and/or by reduc‐
282 ing max_connections to reduce the semaphore consumption.
283 A failure message suggesting that another server is already running
285 should be checked carefully, for example by using the command
286 $ ps ax | grep postgres
288 or
290 $ ps -ef | grep postgres
292 depending on your system. If you are certain that no conflicting server
294 is running, you may remove the lock file mentioned in the message and
295 try again.
296 A failure message indicating inability to bind to a port may indicate
298 that that port is already in use by some non-PostgreSQL process. You
299 may also get this error if you terminate postgres and immediately
300 restart it using the same port; in this case, you must simply wait a
301 few seconds until the operating system closes the port before trying
302 again. Finally, you may get this error if you specify a port number
303 that your operating system considers to be reserved. For example, many
304 versions of Unix consider port numbers under 1024 to be ``trusted'' and
305 only permit the Unix superuser to access them.
306
308 If at all possible, do not use SIGKILL to kill the main postgres
309 server. Doing so will prevent postgres from freeing the system
310 resources (e.g., shared memory and semaphores) that it holds before
311 terminating. This may cause problems for starting a fresh postgres run.
312 To terminate the postgres server normally, the signals SIGTERM, SIGINT,
314 or SIGQUIT can be used. The first will wait for all clients to termi‐
315 nate before quitting, the second will forcefully disconnect all
316 clients, and the third will quit immediately without proper shutdown,
317 resulting in a recovery run during restart. The SIGHUP signal will
318 reload the server configuration files. It is also possible to send
319 SIGHUP to an individual server process, but that is usually not sensi‐
320 ble.
321 The utility command pg_ctl(1) can be used to start and shut down the
323 postgres server safely and comfortably.
324 To cancel a running query, send the SIGINT signal to the process run‐
326 ning that command.
327 The postgres server uses SIGTERM to tell subordinate server processes
329 to quit normally and SIGQUIT to terminate without the normal cleanup.
330 These signals should not be used by users. It is also unwise to send
331 SIGKILL to a server process — the main postgres process will interpret
332 this as a crash and will force all the sibling processes to quit as
333 part of its standard crash-recovery procedure.
334
336 The -- options will not work on FreeBSD or OpenBSD. Use -c instead.
337 This is a bug in the affected operating systems; a future release of
338 PostgreSQL will provide a workaround if this is not fixed.
339
341 To start a single-user mode server, use a command like
342 postgres --single -D /usr/local/pgsql/data other-options my_database
344 Provide the correct path to the database directory with -D, or make
346 sure that the environment variable PGDATA is set. Also specify the
347 name of the particular database you want to work in.
348 Normally, the single-user mode server treats newline as the command
350 entry terminator; there is no intelligence about semicolons, as there
351 is in psql. To continue a command across multiple lines, you must type
352 backslash just before each newline except the last one.
353 But if you use the -j command line switch, then newline does not termi‐
355 nate command entry. In this case, the server will read the standard
356 input until the end-of-file (EOF) marker, then process the input as a
357 single command string. Backslash-newline is not treated specially in
358 this case.
359 To quit the session, type EOF (Control+D, usually). If you've used -j,
361 two consecutive EOFs are needed to exit.
362 Note that the single-user mode server does not provide sophisticated
364 line-editing features (no command history, for example).
365
367 To start postgres in the background using default values, type:
368 $ nohup postgres >logfile 2>&1 </dev/null &
370 To start postgres with a specific port:
373 $ postgres -p 1234
375 This command will start up postgres communicating through the port
377 1234. In order to connect to this server using psql, you would need to
378 run it as
379 $ psql -p 1234
381 or set the environment variable PGPORT:
383 $ export PGPORT=1234
385 $ psql
386 Named run-time parameters can be set in either of these styles:
389 $ postgres -c work_mem=1234
391 $ postgres --work-mem=1234
392 Either form overrides whatever setting might exist for work_mem in
394 postgresql.conf. Notice that underscores in parameter names can be
395 written as either underscore or dash on the command line. Except for
396 short-term experiments, it's probably better practice to edit the set‐
397 ting in postgresql.conf than to rely on a command-line switch to set a
398 parameter.
399
401 initdb(1), pg_ctl(1)
402Application 2008-06-08 POSTGRES(1)