2 Frequently Asked Questions (FAQ) for PostgreSQL
4 Last updated: Wed Apr 16 16:57:09 EDT 2008
6 Current maintainer: Bruce Momjian (bruce@momjian.us)
8 The most recent version of this document can be viewed at
9 http://www.postgresql.org/files/documentation/faqs/FAQ.html.
11 Platform-specific questions are answered at
12 http://www.postgresql.org/docs/faq/.
13 _________________________________________________________________
17 1.1) What is PostgreSQL? How is it pronounced? What is Postgres?
18 1.2) Who controls PostgreSQL?
19 1.3) What is the copyright of PostgreSQL?
20 1.4) What platforms does PostgreSQL support?
21 1.5) Where can I get PostgreSQL?
22 1.6) What is the most recent release?
23 1.7) Where can I get support?
24 1.8) How do I submit a bug report?
25 1.9) How do I find out about known bugs or missing features?
26 1.10) What documentation is available?
27 1.11) How can I learn SQL?
28 1.12) How do I submit a patch or join the development team?
29 1.13) How does PostgreSQL compare to other DBMSs?
30 1.14) Will PostgreSQL handle recent daylight saving time changes in
32 1.15) How do I unsubscribe from the PostgreSQL email lists? How do I
33 avoid receiving duplicate emails?
37 2.1) What interfaces are available for PostgreSQL?
38 2.2) What tools are available for using PostgreSQL with Web pages?
39 2.3) Does PostgreSQL have a graphical user interface?
41 Administrative Questions
43 3.1) How do I install PostgreSQL somewhere other than
45 3.2) How do I control connections from other hosts?
46 3.3) How do I tune the database engine for better performance?
47 3.4) What debugging features are available?
48 3.5) Why do I get "Sorry, too many clients" when trying to connect?
49 3.6 What is the upgrade process for PostgreSQL?
50 3.7) What computer hardware should I use?
54 4.1) How do I SELECT only the first few rows of a query? A random row?
55 4.2) How do I find out what tables, indexes, databases, and users are
56 defined? How do I see the queries used by psql to display them?
57 4.3) How do you change a column's data type?
58 4.4) What is the maximum size for a row, a table, and a database?
59 4.5) How much database disk space is required to store data from a
61 4.6) Why are my queries slow? Why don't they use my indexes?
62 4.7) How do I see how the query optimizer is evaluating my query?
63 4.8) How do I perform regular expression searches and case-insensitive
64 regular expression searches? How do I use an index for
65 case-insensitive searches?
66 4.9) In a query, how do I detect if a field is NULL? How do I
67 concatenate possible NULLs? How can I sort on whether a field is NULL
69 4.10) What is the difference between the various character types?
70 4.11.1) How do I create a serial/auto-incrementing field?
71 4.11.2) How do I get the value of a SERIAL insert?
72 4.11.3) Doesn't currval() lead to a race condition with other users?
73 4.11.4) Why aren't my sequence numbers reused on transaction abort?
74 Why are there gaps in the numbering of my sequence/SERIAL column?
75 4.12) What is an OID? What is a CTID?
76 4.13) Why do I get the error "ERROR: Memory exhausted in
78 4.14) How do I tell what PostgreSQL version I am running?
79 4.15) How do I create a column that will default to the current time?
80 4.16) How do I perform an outer join?
81 4.17) How do I perform queries using multiple databases?
82 4.18) How do I return multiple rows or columns from a function?
83 4.19) Why do I get "relation with OID ##### does not exist" errors
84 when accessing temporary tables in PL/PgSQL functions?
85 4.20) What replication solutions are available?
86 4.21) Why are my table and column names not recognized in my query?
87 Why is capitalization not preserved?
88 _________________________________________________________________
92 1.1) What is PostgreSQL? How is it pronounced? What is Postgres?
94 PostgreSQL is pronounced Post-Gres-Q-L. (For those curious about how
95 to say "PostgreSQL", an audio file is available.)
97 PostgreSQL is an object-relational database system that has the
98 features of traditional commercial database systems with enhancements
99 to be found in next-generation DBMS systems. PostgreSQL is free and
100 the complete source code is available.
102 PostgreSQL development is performed by a team of mostly volunteer
103 developers spread throughout the world and communicating via the
104 Internet. It is a community project and is not controlled by any
105 company. To get involved, see the developer's FAQ at
106 http://www.postgresql.org/docs/faqs.FAQ_DEV.html
108 Postgres is a widely-used nickname for PostgreSQL. It was the original
109 name of the project at Berkeley and is strongly preferred over other
110 nicknames. If you find 'PostgreSQL' hard to pronounce, call it
113 1.2) Who controls PostgreSQL?
115 If you are looking for a PostgreSQL gatekeeper, central committee, or
116 controlling company, give up --- there isn't one. We do have a core
117 committee and CVS committers, but these groups are more for
118 administrative purposes than control. The project is directed by the
119 community of developers and users, which anyone can join. All you need
120 to do is subscribe to the mailing lists and participate in the
121 discussions. (See the Developer's FAQ for information on how to get
122 involved in PostgreSQL development.)
124 1.3) What is the copyright of PostgreSQL?
126 PostgreSQL is distributed under the classic BSD license. Basically, it
127 allows users to do anything they want with the code, including
128 reselling binaries without the source code. The only restriction is
129 that you not hold us legally liable for problems with the software.
130 There is also the requirement that this copyright appear in all copies
131 of the software. Here is the actual BSD license we use:
133 PostgreSQL Data Base Management System
135 Portions Copyright (c) 1996-2008, PostgreSQL Global Development Group
136 Portions Copyright (c) 1994-1996 Regents of the University of
139 Permission to use, copy, modify, and distribute this software and its
140 documentation for any purpose, without fee, and without a written
141 agreement is hereby granted, provided that the above copyright notice
142 and this paragraph and the following two paragraphs appear in all
145 IN NO EVENT SHALL THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY PARTY
146 FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES,
147 INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND
148 ITS DOCUMENTATION, EVEN IF THE UNIVERSITY OF CALIFORNIA HAS BEEN
149 ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
151 THE UNIVERSITY OF CALIFORNIA SPECIFICALLY DISCLAIMS ANY WARRANTIES,
152 INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
153 MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE
154 PROVIDED HEREUNDER IS ON AN "AS IS" BASIS, AND THE UNIVERSITY OF
155 CALIFORNIA HAS NO OBLIGATIONS TO PROVIDE MAINTENANCE, SUPPORT,
156 UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
158 1.4) What platforms does PostgreSQL support?
160 In general, any modern Unix-compatible platform should be able to run
161 PostgreSQL. The platforms that had received explicit testing at the
162 time of release are listed in the installation instructions.
164 PostgreSQL also runs natively on Microsoft Windows NT-based operating
165 systems like Win2000 SP4, WinXP, and Win2003. A prepackaged installer
166 is available at http://pgfoundry.org/projects/pginstaller. MSDOS-based
167 versions of Windows (Win95, Win98, WinMe) can run PostgreSQL using
170 There is also a Novell Netware 6 port at http://forge.novell.com, and
171 an OS/2 (eComStation) version at
172 http://hobbes.nmsu.edu/cgi-bin/h-search?sh=1&button=Search&key=postgre
173 SQL&stype=all&sort=type&dir=%2F.
175 1.5) Where can I get PostgreSQL?
177 Via web browser, use http://www.postgresql.org/ftp/, and via ftp, use
178 ftp://ftp.postgresql.org/pub/.
180 1.6) What is the most recent release?
182 The latest release of PostgreSQL is version 8.3.1.
184 We plan to have a major release every year, with minor releases every
187 1.7) Where can I get support?
189 The PostgreSQL community provides assistance to many of its users via
190 email. The main web site to subscribe to the email lists is
191 http://www.postgresql.org/community/lists/. The general or bugs lists
192 are a good place to start.
194 The major IRC channel is #postgresql on Freenode (irc.freenode.net).
195 To connect you can use the Unix program irc -c '#postgresql' "$USER"
196 irc.freenode.net or use any other IRC clients. A Spanish one also
197 exists on the same network, (#postgresql-es), a French one,
198 (#postgresqlfr), and a Brazilian one, (#postgresql-br). There is also
199 a PostgreSQL channel on EFNet.
201 A list of commercial support companies is available at
202 http://www.postgresql.org/support/professional_support.
204 1.8) How do I submit a bug report?
206 Visit the PostgreSQL bug form at
207 http://www.postgresql.org/support/submitbug. Also check out our ftp
208 site ftp://ftp.postgresql.org/pub/ to see if there is a more recent
211 Bugs submitted using the bug form or posted to any PostgreSQL mailing
212 list typically generates one of the following replies:
213 * It is not a bug, and why
214 * It is a known bug and is already on the TODO list
215 * The bug has been fixed in the current release
216 * The bug has been fixed but is not packaged yet in an official
218 * A request is made for more detailed information:
221 + Reproducible test case
222 + Debugging information
223 + Debugger backtrace output
224 * The bug is new. The following might happen:
225 + A patch is created and will be included in the next major or
227 + The bug cannot be fixed immediately and is added to the TODO
230 1.9) How do I find out about known bugs or missing features?
232 PostgreSQL supports an extended subset of SQL:2003. See our TODO list
233 for known bugs, missing features, and future plans.
235 A feature request usually results in one of the following replies:
236 * The feature is already on the TODO list
237 * The feature is not desired because:
238 + It duplicates existing functionality that already follows the
240 + The feature would increase code complexity but add little
242 + The feature would be insecure or unreliable
243 * The new feature is added to the TODO list
245 PostgreSQL does not use a bug tracking system because we find it more
246 efficient to respond directly to email and keep the TODO list
247 up-to-date. In practice, bugs don't last very long in the software,
248 and bugs that affect a large number of users are fixed rapidly. The
249 only place to find all changes, improvements, and fixes in a
250 PostgreSQL release is to read the CVS log messages. Even the release
251 notes do not list every change made to the software.
253 1.10) What documentation is available?
255 PostgreSQL includes extensive documentation, including a large manual,
256 manual pages, and some test examples. See the /doc directory. You can
257 also browse the manuals online at http://www.postgresql.org/docs.
259 There are two PostgreSQL books available online at
260 http://www.postgresql.org/docs/books/awbook.html and
261 http://www.commandprompt.com/ppbook/. There are a number of PostgreSQL
262 books available for purchase. One of the most popular ones is by Korry
263 Douglas. A list of book reviews can be found at
264 http://www.postgresql.org/docs/books/. There is also a collection of
265 PostgreSQL technical articles at
266 http://www.postgresql.org/docs/techdocs.
268 The command line client program psql has some \d commands to show
269 information about types, operators, functions, aggregates, etc. - use
270 \? to display the available commands.
272 Our web site contains even more documentation.
274 1.11) How can I learn SQL?
276 First, consider the PostgreSQL-specific books mentioned above. Many of
277 our users also like The Practical SQL Handbook, Bowman, Judith S., et
278 al., Addison-Wesley. Others like The Complete Reference SQL, Groff et
281 There are also many nice tutorials available online:
282 * http://www.intermedia.net/support/sql/sqltut.shtm
283 * http://sqlcourse.com
284 * http://www.w3schools.com/sql/default.asp
285 * http://mysite.verizon.net/Graeme_Birchall/id1.html
287 1.12) How do I submit a patch or join the development team?
289 See the Developer's FAQ.
291 1.13) How does PostgreSQL compare to other DBMSs?
293 There are several ways of measuring software: features, performance,
294 reliability, support, and price.
297 PostgreSQL has most features present in large commercial DBMSs,
298 like transactions, subselects, triggers, views, foreign key
299 referential integrity, and sophisticated locking. We have some
300 features they do not have, like user-defined types,
301 inheritance, rules, and multi-version concurrency control to
302 reduce lock contention.
305 PostgreSQL's performance is comparable to other commercial and
306 open source databases. It is faster for some things, slower for
307 others. Our performance is usually +/-10% compared to other
311 We realize that a DBMS must be reliable, or it is worthless. We
312 strive to release well-tested, stable code that has a minimum
313 of bugs. Each release has at least one month of beta testing,
314 and our release history shows that we can provide stable, solid
315 releases that are ready for production use. We believe we
316 compare favorably to other database software in this area.
319 Our mailing lists provide contact with a large group of
320 developers and users to help resolve any problems encountered.
321 While we cannot guarantee a fix, commercial DBMSs do not always
322 supply a fix either. Direct access to developers, the user
323 community, manuals, and the source code often make PostgreSQL
324 support superior to other DBMSs. There is commercial
325 per-incident support available for those who need it. (See FAQ
329 We are free for all use, both commercial and non-commercial.
330 You can add our code to your product with no limitations,
331 except those outlined in our BSD-style license stated above.
333 1.14) Will PostgreSQL handle recent daylight saving time changes in various
336 USA daylight saving time changes are included in PostgreSQL release
337 8.0.[4+], and all later major releases, e.g. 8.1. Canada and Western
338 Australia changes are included in 8.0.[10+], 8.1.[6+], and all later
339 major releases. PostgreSQL releases prior to 8.0 use the operating
340 system's timezone database for daylight saving information.
342 1.15) How do I unsubscribe from the PostgreSQL email lists? How do I avoid
343 receiving duplicate emails?
345 The PostgreSQL Majordomo page allows subscribing or unsubscribing from
346 any of the PostgreSQL email lists. (You might need to have your
347 Majordomo password emailed to you to log in.)
349 All PostgreSQL email lists are configured so a group reply goes to the
350 email list and the original email author. This is done so users
351 receive the quickest possible email replies. If you would prefer not
352 to receive duplicate email from the list in cases where you already
353 receive an email directly, check eliminatecc from the Majordomo Change
354 Settings page. You can also prevent yourself from receiving copies of
355 emails you post to the lists by unchecking selfcopy.
356 _________________________________________________________________
358 User Client Questions
360 2.1) What interfaces are available for PostgreSQL?
362 The PostgreSQL install includes only the C and embedded C interfaces.
363 All other interfaces are independent projects that are downloaded
364 separately; being separate allows them to have their own release
365 schedule and development teams.
367 Some programming languages like PHP include an interface to
368 PostgreSQL. Interfaces for languages like Perl, TCL, Python, and many
369 others are available at http://pgfoundry.org.
371 2.2) What tools are available for using PostgreSQL with Web pages?
373 A nice introduction to Database-backed Web pages can be seen at:
374 http://www.webreview.com
376 For Web integration, PHP (http://www.php.net) is an excellent
379 For complex cases, many use the Perl and DBD::Pg with CGI.pm or
382 2.3) Does PostgreSQL have a graphical user interface?
384 There are a large number of GUI Tools that are available for
385 PostgreSQL from both commercial and open source developers. A detailed
386 list can be found in the PostgreSQL Community Documentation
387 _________________________________________________________________
389 Administrative Questions
391 3.1) How do I install PostgreSQL somewhere other than /usr/local/pgsql?
393 Specify the --prefix option when running configure.
395 3.2) How do I control connections from other hosts?
397 By default, PostgreSQL only allows connections from the local machine
398 using Unix domain sockets or TCP/IP connections. Other machines will
399 not be able to connect unless you modify listen_addresses in the
400 postgresql.conf file, enable host-based authentication by modifying
401 the $PGDATA/pg_hba.conf file, and restart the database server.
403 3.3) How do I tune the database engine for better performance?
405 There are three major areas for potential performance improvement:
408 This involves modifying queries to obtain better performance:
410 + Creation of indexes, including expression and partial indexes
411 + Use of COPY instead of multiple INSERTs
412 + Grouping of multiple statements into a single transaction to
413 reduce commit overhead
414 + Use of CLUSTER when retrieving many rows from an index
415 + Use of LIMIT for returning a subset of a query's output
416 + Use of Prepared queries
417 + Use of ANALYZE to maintain accurate optimizer statistics
418 + Regular use of VACUUM or pg_autovacuum
419 + Dropping of indexes during large data changes
422 A number of postgresql.conf settings affect performance. For
423 more details, see Administration Guide/Server Run-time
424 Environment/Run-time Configuration for a full listing, and for
426 http://www.varlena.com/varlena/GeneralBits/Tidbits/annotated_co
428 http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html.
431 The effect of hardware on performance is detailed in
432 http://www.powerpostgresql.com/PerfList/ and
433 http://momjian.us/main/writings/pgsql/hw_performance/index.html
436 3.4) What debugging features are available?
438 There are many log_* server configuration variables that enable
439 printing of query and process statistics which can be very useful for
440 debugging and performance measurements.
442 3.5) Why do I get "Sorry, too many clients" when trying to connect?
444 You have reached the default limit of 100 database sessions. You need
445 to increase the server's limit on how many concurrent backend
446 processes it can start by changing the max_connections value in
447 postgresql.conf and restarting the server.
449 3.6) What is the upgrade process for PostgreSQL?
451 See http://www.postgresql.org/support/versioning for a general
452 discussion about upgrading, and
453 http://www.postgresql.org/docs/current/static/install-upgrading.html
454 for specific instructions.
456 3.7) What computer hardware should I use?
458 Because PC hardware is mostly compatible, people tend to believe that
459 all PC hardware is of equal quality. It is not. ECC RAM, SCSI, and
460 quality motherboards are more reliable and have better performance
461 than less expensive hardware. PostgreSQL will run on almost any
462 hardware, but if reliability and performance are important it is wise
463 to research your hardware options thoroughly. Our email lists can be
464 used to discuss hardware options and tradeoffs.
465 _________________________________________________________________
467 Operational Questions
469 4.1) How do I SELECT only the first few rows of a query? A random row?
471 To retrieve only a few rows, if you know at the number of rows needed
472 at the time of the SELECT use LIMIT . If an index matches the ORDER BY
473 it is possible the entire query does not have to be executed. If you
474 don't know the number of rows at SELECT time, use a cursor and FETCH.
476 To SELECT a random row, use:
482 4.2) How do I find out what tables, indexes, databases, and users are
483 defined? How do I see the queries used by psql to display them?
485 Use the \dt command to see tables in psql. For a complete list of
486 commands inside psql you can use \?. Alternatively you can read the
487 source code for psql in file pgsql/src/bin/psql/describe.c, it
488 contains SQL commands that generate the output for psql's backslash
489 commands. You can also start psql with the -E option so it will print
490 out the queries it uses to execute the commands you give. PostgreSQL
491 also provides an SQL compliant INFORMATION SCHEMA interface you can
492 query to get information about the database.
494 There are also system tables beginning with pg_ that describe these
497 Use psql -l will list all databases.
499 Also try the file pgsql/src/tutorial/syscat.source. It illustrates
500 many of the SELECTs needed to get information from the database system
503 4.3) How do you change a column's data type?
505 Changing the data type of a column can be done easily in 8.0 and later
506 with ALTER TABLE ALTER COLUMN TYPE.
508 In earlier releases, do this:
510 ALTER TABLE tab ADD COLUMN new_col new_data_type;
511 UPDATE tab SET new_col = CAST(old_col AS new_data_type);
512 ALTER TABLE tab DROP COLUMN old_col;
515 You might then want to do VACUUM FULL tab to reclaim the disk space
516 used by the expired rows.
518 4.4) What is the maximum size for a row, a table, and a database?
520 These are the limits:
522 Maximum size for a database? unlimited (32 TB databases exist)
523 Maximum size for a table? 32 TB
524 Maximum size for a row? 400 GB
525 Maximum size for a field? 1 GB
526 Maximum number of rows in a table? unlimited
527 Maximum number of columns in a table? 250-1600 depending on column
529 Maximum number of indexes on a table? unlimited
531 Of course, these are not actually unlimited, but limited to available
532 disk space and memory/swap space. Performance may suffer when these
533 values get unusually large.
535 The maximum table size of 32 TB does not require large file support
536 from the operating system. Large tables are stored as multiple 1 GB
537 files so file system size limits are not important.
539 The maximum table size, row size, and maximum number of columns can be
540 quadrupled by increasing the default block size to 32k. The maximum
541 table size can also be increased using table partitioning.
543 One limitation is that indexes can not be created on columns longer
544 than about 2,000 characters. Fortunately, such indexes are rarely
545 needed. Uniqueness is best guaranteed by a function index of an MD5
546 hash of the long column, and full text indexing allows for searching
547 of words within the column.
549 4.5) How much database disk space is required to store data from a typical
552 A PostgreSQL database may require up to five times the disk space to
553 store data from a text file.
555 As an example, consider a file of 100,000 lines with an integer and
556 text description on each line. Suppose the text string avergages
557 twenty bytes in length. The flat file would be 2.8 MB. The size of the
558 PostgreSQL database file containing this data can be estimated as 5.2
560 24 bytes: each row header (approximate)
561 24 bytes: one int field and one text field
562 + 4 bytes: pointer on page to tuple
563 ----------------------------------------
566 The data page size in PostgreSQL is 8192 bytes (8 KB), so:
569 ------------------- = 158 rows per database page (rounded down)
573 -------------------- = 633 database pages (rounded up)
576 633 database pages * 8192 bytes per page = 5,185,536 bytes (5.2 MB)
578 Indexes do not require as much overhead, but do contain the data that
579 is being indexed, so they can be large also.
581 NULLs are stored as bitmaps, so they use very little space.
583 4.6) Why are my queries slow? Why don't they use my indexes?
585 Indexes are not used by every query. Indexes are used only if the
586 table is larger than a minimum size, and the query selects only a
587 small percentage of the rows in the table. This is because the random
588 disk access caused by an index scan can be slower than a straight read
589 through the table, or sequential scan.
591 To determine if an index should be used, PostgreSQL must have
592 statistics about the table. These statistics are collected using
593 VACUUM ANALYZE, or simply ANALYZE. Using statistics, the optimizer
594 knows how many rows are in the table, and can better determine if
595 indexes should be used. Statistics are also valuable in determining
596 optimal join order and join methods. Statistics collection should be
597 performed periodically as the contents of the table change.
599 Indexes are normally not used for ORDER BY or to perform joins. A
600 sequential scan followed by an explicit sort is usually faster than an
601 index scan of a large table. However, LIMIT combined with ORDER BY
602 often will use an index because only a small portion of the table is
605 If you believe the optimizer is incorrect in choosing a sequential
606 scan, use SET enable_seqscan TO 'off' and run query again to see if an
607 index scan is indeed faster.
609 When using wild-card operators such as LIKE or ~, indexes can only be
610 used in certain circumstances:
611 * The beginning of the search string must be anchored to the start
613 + LIKE patterns must not start with %.
614 + ~ (regular expression) patterns must start with ^.
615 * The search string can not start with a character class, e.g.
617 * Case-insensitive searches such as ILIKE and ~* do not utilize
618 indexes. Instead, use expression indexes, which are described in
620 * The default C locale must be used during initdb because it is not
621 possible to know the next-greatest character in a non-C locale.
622 You can create a special text_pattern_ops index for such cases
623 that work only for LIKE indexing. It is also possible to use full
624 text indexing for word searches.
626 4.7) How do I see how the query optimizer is evaluating my query?
628 See the EXPLAIN manual page.
630 4.8) How do I perform regular expression searches and case-insensitive
631 regular expression searches? How do I use an index for case-insensitive
634 The ~ operator does regular expression matching, and ~* does
635 case-insensitive regular expression matching. The case-insensitive
636 variant of LIKE is called ILIKE.
638 Case-insensitive equality comparisons are normally expressed as:
641 WHERE lower(col) = 'abc';
643 This will not use an standard index. However, if you create an
644 expression index, it will be used:
645 CREATE INDEX tabindex ON tab (lower(col));
647 If the above index is created as UNIQUE, though the column can store
648 upper and lowercase characters, it can not have identical values that
649 differ only in case. To force a particular case to be stored in the
650 column, use a CHECK constraint or a trigger.
652 4.9) In a query, how do I detect if a field is NULL? How do I concatenate
653 possible NULLs? How can I sort on whether a field is NULL or not?
655 You test the column with IS NULL and IS NOT NULL, like this:
660 To concatentate with possible NULLs, use COALESCE(), like this:
661 SELECT COALESCE(col1, '') || COALESCE(col2, '')
664 To sort by the NULL status, use the IS NULL and IS NOT NULL modifiers
665 in your ORDER BY clause. Things that are true will sort higher than
666 things that are false, so the following will put NULL entries at the
667 top of the resulting list:
670 ORDER BY (col IS NOT NULL)
672 4.10) What is the difference between the various character types?
674 Type Internal Name Notes
675 VARCHAR(n) varchar size specifies maximum length, no padding
676 CHAR(n) bpchar blank padded to the specified fixed length
677 TEXT text no specific upper limit on length
678 BYTEA bytea variable-length byte array (null-byte safe)
679 "char" char one character
681 You will see the internal name when examining system catalogs and in
684 The first four types above are "varlena" types (i.e., the first four
685 bytes on disk are the length, followed by the data). Thus the actual
686 space used is slightly greater than the declared size. However, long
687 values are also subject to compression, so the space on disk might
688 also be less than expected.
689 VARCHAR(n) is best when storing variable-length strings and it limits
690 how long a string can be. TEXT is for strings of unlimited length,
691 with a maximum of one gigabyte.
693 CHAR(n) is for storing strings that are all the same length. CHAR(n)
694 pads with blanks to the specified length, while VARCHAR(n) only stores
695 the characters supplied. BYTEA is for storing binary data,
696 particularly values that include NULL bytes. All the types described
697 here have similar performance characteristics.
699 4.11.1) How do I create a serial/auto-incrementing field?
701 PostgreSQL supports a SERIAL data type. It auto-creates a sequence.
703 CREATE TABLE person (
708 is automatically translated into this:
709 CREATE SEQUENCE person_id_seq;
710 CREATE TABLE person (
711 id INT4 NOT NULL DEFAULT nextval('person_id_seq'),
715 Automatically created sequence are named <table>_<serialcolumn>_seq,
716 where table and serialcolumn are the names of the table and SERIAL
717 column, respectively. See the create_sequence manual page for more
718 information about sequences.
720 4.11.2) How do I get the value of a SERIAL insert?
722 The simplest way is to retrieve the assigned SERIAL value with
723 RETURNING. Using the example table in 4.11.1, it would look like this:
724 INSERT INTO person (name) VALUES ('Blaise Pascal') RETURNING id;
726 You can also call nextval() and use that value in the INSERT, or call
727 currval() after the INSERT.
729 4.11.3) Doesn't currval() lead to a race condition with other users?
731 No. currval() returns the current value assigned by your session, not
734 4.11.4) Why aren't my sequence numbers reused on transaction abort? Why are
735 there gaps in the numbering of my sequence/SERIAL column?
737 To improve concurrency, sequence values are given out to running
738 transactions as needed and are not locked until the transaction
739 completes. This causes gaps in numbering from aborted transactions.
741 4.12) What is an OID? What is a CTID?
743 If a table is created WITH OIDS, each row gets a unique a OID. OIDs
744 are automatically assigned unique 4-byte integers that are unique
745 across the entire installation. However, they overflow at 4 billion,
746 and then the OIDs start being duplicated. PostgreSQL uses OIDs to link
747 its internal system tables together.
749 To uniquely number rows in user tables, it is best to use SERIAL
750 rather than OIDs because SERIAL sequences are unique only within a
751 single table. and are therefore less likely to overflow. SERIAL8 is
752 available for storing eight-byte sequence values.
754 CTIDs are used to identify specific physical rows with block and
755 offset values. CTIDs change after rows are modified or reloaded. They
756 are used by index entries to point to physical rows.
758 4.13) Why do I get the error "ERROR: Memory exhausted in AllocSetAlloc()"?
760 You probably have run out of virtual memory on your system, or your
761 kernel has a low limit for certain resources. Try this before starting
766 Depending on your shell, only one of these may succeed, but it will
767 set your process data segment limit much higher and perhaps allow the
768 query to complete. This command applies to the current process, and
769 all subprocesses created after the command is run. If you are having a
770 problem with the SQL client because the backend is returning too much
771 data, try it before starting the client.
773 4.14) How do I tell what PostgreSQL version I am running?
775 From psql, type SELECT version();
777 4.15) How do I create a column that will default to the current time?
779 Use CURRENT_TIMESTAMP:
780 CREATE TABLE test (x int, modtime TIMESTAMP DEFAULT CURRENT_TIMESTAMP );
782 4.16) How do I perform an outer join?
784 PostgreSQL supports outer joins using the SQL standard syntax. Here
787 FROM t1 LEFT OUTER JOIN t2 ON (t1.col = t2.col);
791 FROM t1 LEFT OUTER JOIN t2 USING (col);
793 These identical queries join t1.col to t2.col, and also return any
794 unjoined rows in t1 (those with no match in t2). A RIGHT join would
795 add unjoined rows of t2. A FULL join would return the matched rows
796 plus all unjoined rows from t1 and t2. The word OUTER is optional and
797 is assumed in LEFT, RIGHT, and FULL joins. Ordinary joins are called
800 4.17) How do I perform queries using multiple databases?
802 There is no way to query a database other than the current one.
803 Because PostgreSQL loads database-specific system catalogs, it is
804 uncertain how a cross-database query should even behave.
806 contrib/dblink allows cross-database queries using function calls. Of
807 course, a client can also make simultaneous connections to different
808 databases and merge the results on the client side.
810 4.18) How do I return multiple rows or columns from a function?
812 It is easy using set-returning functions,
813 http://www.postgresql.org/docs/techdocs.17.
815 4.19) Why do I get "relation with OID ##### does not exist" errors when
816 accessing temporary tables in PL/PgSQL functions?
818 In PostgreSQL versions < 8.3, PL/PgSQL caches function scripts, and an
819 unfortunate side effect is that if a PL/PgSQL function accesses a
820 temporary table, and that table is later dropped and recreated, and
821 the function called again, the function will fail because the cached
822 function contents still point to the old temporary table. The solution
823 is to use EXECUTE for temporary table access in PL/PgSQL. This will
824 cause the query to be reparsed every time.
826 This problem does not occur in PostgreSQL 8.3 and later.
828 4.20) What replication solutions are available?
830 Though "replication" is a single term, there are several technologies
831 for doing replication, with advantages and disadvantages for each.
833 Master/slave replication allows a single master to receive read/write
834 queries, while slaves can only accept read/SELECT queries. The most
835 popular freely available master-slave PostgreSQL replication solution
838 Multi-master replication allows read/write queries to be sent to
839 multiple replicated computers. This capability also has a severe
840 impact on performance due to the need to synchronize changes between
841 servers. PGCluster is the most popular such solution freely available
844 There are also commercial and hardware-based replication solutions
845 available supporting a variety of replication models.
847 4.21) Why are my table and column names not recognized in my query? Why is
848 capitalization not preserved?
850 The most common cause of unrecognized names is the use of
851 double-quotes around table or column names during table creation. When
852 double-quotes are used, table and column names (called identifiers)
853 are stored case-sensitive, meaning you must use double-quotes when
854 referencing the names in a query. Some interfaces, like pgAdmin,
855 automatically double-quote identifiers during table creation. So, for
856 identifiers to be recognized, you must either:
857 * Avoid double-quoting identifiers when creating tables
858 * Use only lowercase characters in identifiers
859 * Double-quote identifiers when referencing them in queries