1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
4 <META name="generator" content="HTML Tidy, see www.w3.org">
5 <META http-equiv="Content-Type" content="text/html; charset=US-ASCII">
6 <TITLE>PostgreSQL FAQ</TITLE>
9 <BODY bgcolor="#ffffff" text="#000000" link="#ff0000" vlink="#a00000"
11 <H1>Frequently Asked Questions (FAQ) for PostgreSQL</H1>
13 <P>Last updated: Fri Oct 3 21:39:47 EDT 2003</P>
15 <P>Current maintainer: Bruce Momjian (<A href=
16 "mailto:pgman@candle.pha.pa.us">pgman@candle.pha.pa.us</A>)<BR>
19 <P>The most recent version of this document can be viewed at <A
21 "http://www.PostgreSQL.org/docs/faqs/FAQ.html">http://www.PostgreSQL.org/docs/faqs/FAQ.html</A>.</P>
23 <P>Platform-specific questions are answered at <A href=
24 "http://www.PostgreSQL.org/docs/index.html">http://www.PostgreSQL.org/docs/index.html</A>.</P>
27 <H2 align="center">General Questions</H2>
28 <A href="#1.1">1.1</A>) What is PostgreSQL? How is it pronounced?<BR>
29 <A href="#1.2">1.2</A>) What is the copyright on PostgreSQL?<BR>
30 <A href="#1.3">1.3</A>) What Unix platforms does PostgreSQL run
32 <A href="#1.4">1.4</A>) What non-Unix ports are available?<BR>
33 <A href="#1.5">1.5</A>) Where can I get PostgreSQL?<BR>
34 <A href="#1.6">1.6</A>) Where can I get support?<BR>
35 <A href="#1.7">1.7</A>) What is the latest release?<BR>
36 <A href="#1.8">1.8</A>) What documentation is available?<BR>
37 <A href="#1.9">1.9</A>) How do I find out about known bugs or
39 <A href="#1.10">1.10</A>) How can I learn <SMALL>SQL</SMALL>?<BR>
40 <A href="#1.11">1.11</A>) Is PostgreSQL Y2K compliant?<BR>
41 <A href="#1.12">1.12</A>) How do I join the development team?<BR>
42 <A href="#1.13">1.13</A>) How do I submit a bug report?<BR>
43 <A href="#1.14">1.14</A>) How does PostgreSQL compare to other
44 <SMALL>DBMS</SMALL>s?<BR>
45 <A href="#1.15">1.15</A>) How can I financially assist
49 <H2 align="center">User Client Questions</H2>
50 <A href="#2.1">2.1</A>) Are there <SMALL>ODBC</SMALL> drivers for
52 <A href="#2.2">2.2</A>) What tools are available for using
53 PostgreSQL with Web pages?<BR>
54 <A href="#2.3">2.3</A>) Does PostgreSQL have a graphical user
56 <A href="#2.4">2.4</A>) What languages are available to
57 communicate with PostgreSQL?<BR>
60 <H2 align="center">Administrative Questions</H2>
61 <A href="#3.1">3.1</A>) How do I install PostgreSQL somewhere other
62 than <I>/usr/local/pgsql</I>?<BR>
63 <A href="#3.2">3.2</A>) When I start <I>postmaster</I>, I get a
64 <I>Bad System Call</I> or core dumped message. Why?<BR>
65 <A href="#3.3">3.3</A>) When I try to start <I>postmaster</I>, I
66 get <I>IpcMemoryCreate</I> errors. Why?<BR>
67 <A href="#3.4">3.4</A>) When I try to start <I>postmaster</I>, I
68 get <I>IpcSemaphoreCreate</I> errors. Why?<BR>
69 <A href="#3.5">3.5</A>) How do I control connections from other
71 <A href="#3.6">3.6</A>) How do I tune the database engine for
72 better performance?<BR>
73 <A href="#3.7">3.7</A>) What debugging features are available?<BR>
74 <A href="#3.8">3.8</A>) Why do I get <I>"Sorry, too many
75 clients"</I> when trying to connect?<BR>
76 <A href="#3.9">3.9</A>) What is in the <I>pgsql_tmp</I>
78 <A href="#3.10">3.10</A>) Why do I need to do a dump and restore
79 to upgrade PostgreSQL releases?<BR>
82 <H2 align="center">Operational Questions</H2>
83 <A href="#4.1">4.1</A>) What is the difference between binary
84 cursors and normal cursors?<BR>
85 <A href="#4.2">4.2</A>) How do I <SMALL>SELECT</SMALL> only the
86 first few rows of a query? A random row?<BR>
87 <A href="#4.3">4.3</A>) How do I get a list of tables or other
88 things I can see in <I>psql</I>?<BR>
89 <A href="#4.4">4.4</A>) How do you remove a column from a
90 table, or change it's data type?<BR>
91 <A href="#4.5">4.5</A>) What is the maximum size for a row, a
92 table, and a database?<BR>
93 <A href="#4.6">4.6</A>) How much database disk space is required
94 to store data from a typical text file?<BR>
95 <A href="#4.7">4.7</A>) How do I find out what tables, indexes,
96 databases, and users are defined?<BR>
97 <A href="#4.8">4.8</A>) My queries are slow or don't make use of
99 <A href="#4.9">4.9</A>) How do I see how the query optimizer is
100 evaluating my query?<BR>
101 <A href="#4.10">4.10</A>) What is an R-tree index?<BR>
102 <A href="#4.11">4.11</A>) What is the Genetic Query Optimizer?<BR>
103 <A href="#4.12">4.12</A>) How do I perform regular expression
104 searches and case-insensitive regular expression searches? How do I
105 use an index for case-insensitive searches?<BR>
106 <A href="#4.13">4.13</A>) In a query, how do I detect if a field
107 is <SMALL>NULL</SMALL>?<BR>
108 <A href="#4.14">4.14</A>) What is the difference between the
109 various character types?<BR>
110 <A href="#4.15.1">4.15.1</A>) How do I create a
111 serial/auto-incrementing field?<BR>
112 <A href="#4.15.2">4.15.2</A>) How do I get the value of a
113 <SMALL>SERIAL</SMALL> insert?<BR>
114 <A href="#4.15.3">4.15.3</A>) Don't <I>currval()</I> and
115 <I>nextval()</I> lead to a race condition with other users?<BR>
116 <A href="#4.15.4">4.15.4</A>) Why aren't my sequence numbers
117 reused on transaction abort? Why are there gaps in the numbering of
118 my sequence/SERIAL column?<BR>
119 <A href="#4.16">4.16</A>) What is an <SMALL>OID</SMALL>? What is a
120 <SMALL>TID</SMALL>?<BR>
121 <A href="#4.17">4.17</A>) What is the meaning of some of the terms
122 used in PostgreSQL?<BR>
123 <A href="#4.18">4.18</A>) Why do I get the error <I>"ERROR: Memory
124 exhausted in AllocSetAlloc()"</I>?<BR>
125 <A href="#4.19">4.19</A>) How do I tell what PostgreSQL version I
127 <A href="#4.20">4.20</A>) Why does my large-object operations get
128 <I>"invalid large obj descriptor"</I>?<BR>
129 <A href="#4.21">4.21</A>) How do I create a column that will
130 default to the current time?<BR>
131 <A href="#4.22">4.22</A>) Why are my subqueries using
132 <CODE><SMALL>IN</SMALL></CODE> so slow?<BR>
133 <A href="#4.23">4.23</A>) How do I perform an outer join?<BR>
134 <A href="#4.24">4.24</A>) How do I perform queries using multiple
136 <A href="#4.25">4.25</A>) How do I return multiple rows or columns
138 <A href="#4.26">4.26</A>) Why can't I reliably create/drop
139 temporary tables in PL/PgSQL functions?<BR>
140 <A href="#4.27">4.27</A>) What replication options are available?<BR>
141 <A href="#4.28">4.28</A>) What encryption options are available?<BR>
144 <H2 align="center">Extending PostgreSQL</H2>
145 <A href="#5.1">5.1</A>) I wrote a user-defined function. When I run
146 it in <I>psql</I>, why does it dump core?<BR>
147 <A href="#5.2">5.2</A>) How can I contribute some nifty new types
148 and functions to PostgreSQL?<BR>
149 <A href="#5.3">5.3</A>) How do I write a C function to return a
151 <A href="#5.4">5.4</A>) I have changed a source file. Why does the
152 recompile not see the change?<BR>
156 <H2 align="center">General Questions</H2>
158 <H4><A name="1.1">1.1</A>) What is PostgreSQL? How is it pronounced?</H4>
160 <P>PostgreSQL is pronounced <I>Post-Gres-Q-L</I>.</P>
162 <P>PostgreSQL is an enhancement of the POSTGRES database management
163 system, a next-generation <SMALL>DBMS</SMALL> research prototype.
164 While PostgreSQL retains the powerful data model and rich data
165 types of POSTGRES, it replaces the PostQuel query language with an
166 extended subset of <SMALL>SQL</SMALL>. PostgreSQL is free and the
167 complete source is available.</P>
169 <P>PostgreSQL development is performed by a team of
170 developers who all subscribe to the PostgreSQL development mailing
171 list. The current coordinator is Marc G. Fournier (<A href=
172 "mailto:scrappy@PostgreSQL.org">scrappy@PostgreSQL.org</A>). (See
173 section <a href="#1.6">1.6</a> on how to join). This team is now
174 responsible for all development of PostgreSQL.</P>
176 <P>The authors of PostgreSQL 1.01 were Andrew Yu and Jolly Chen.
177 Many others have contributed to the porting, testing, debugging,
178 and enhancement of the code. The original Postgres code, from which
179 PostgreSQL is derived, was the effort of many graduate students,
180 undergraduate students, and staff programmers working under the
181 direction of Professor Michael Stonebraker at the University of
182 California, Berkeley.</P>
184 <P>The original name of the software at Berkeley was Postgres. When
185 <SMALL>SQL</SMALL> functionality was added in 1995, its name was
186 changed to Postgres95. The name was changed at the end of 1996 to
189 <H4><A name="1.2">1.2</A>) What is the copyright on
192 <P>PostgreSQL is subject to the following COPYRIGHT:</P>
194 <P>PostgreSQL Data Base Management System</P>
196 <P>Portions copyright (c) 1996-2002, PostgreSQL Global Development
197 Group Portions Copyright (c) 1994-6 Regents of the University of
200 <P>Permission to use, copy, modify, and distribute this software
201 and its documentation for any purpose, without fee, and without a
202 written agreement is hereby granted, provided that the above
203 copyright notice and this paragraph and the following two
204 paragraphs appear in all copies.</P>
206 <P>IN NO EVENT SHALL THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY
207 PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL
208 DAMAGES, INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF THIS
209 SOFTWARE AND ITS DOCUMENTATION, EVEN IF THE UNIVERSITY OF
210 CALIFORNIA HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.</P>
212 <P>THE UNIVERSITY OF CALIFORNIA SPECIFICALLY DISCLAIMS ANY
213 WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
214 OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE
215 SOFTWARE PROVIDED HEREUNDER IS ON AN "AS IS" BASIS, AND THE
216 UNIVERSITY OF CALIFORNIA HAS NO OBLIGATIONS TO PROVIDE MAINTENANCE,
217 SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.</P>
219 <P>The above is the BSD license, the classic open-source license.
220 It has no restrictions on how the source code may be used. We like
221 it and have no intention of changing it.</P>
223 <H4><A name="1.3">1.3</A>) What Unix platforms does PostgreSQL run
226 <P>In general, a modern Unix-compatible platform should be able to
227 run PostgreSQL. The platforms that had received explicit testing at
228 the time of release are listed in the installation
231 <H4><A name="1.4">1.4</A>) What non-Unix ports are available?</H4>
233 <P><STRONG>Client</STRONG></P>
235 <P>It is possible to compile the <I>libpq</I> C library, psql, and
236 other interfaces and client applications to run on MS Windows platforms.
237 In this case, the client is running on MS Windows, and communicates
238 via TCP/IP to a server running on one of our supported Unix
239 platforms. A file <I>win32.mak</I> is included in the distribution
240 for making a Win32 <I>libpq</I> library and <I>psql</I>. PostgreSQL
241 also communicates with <SMALL>ODBC</SMALL> clients.</P>
243 <P><STRONG>Server</STRONG></P>
245 <P>The database server can run on Windows NT and Win2k using
246 Cygwin, the Cygnus Unix/NT porting library. See
247 <I>pgsql/doc/FAQ_MSWIN</I> in the distribution or the MS Windows FAQ
248 at <A href="http://www.PostgreSQL.org/docs/faqs/text/FAQ_MSWIN">
249 http://www.PostgreSQL.org/docs/faqs/text/FAQ_MSWIN</A>.</P>
251 <p>A native port to MS Win NT/2000/XP is currently being worked
252 on. For more details on the current status of PostgreSQL on Windows see
253 <a href="http://techdocs.postgresql.org/guides/Windows">
254 http://techdocs.postgresql.org/guides/Windows</a> and
255 <a href="http://momjian.postgresql.org/main/writings/pgsql/win32.html">
256 http://momjian.postgresql.org/main/writings/pgsql/win32.html</a>.</p>
258 <p>There is also a Novell Netware 6 port at
259 <a href="http://forge.novell.com">http://forge.novell.com</a>.</p>
261 <H4><A name="1.5">1.5</A>) Where can I get PostgreSQL?</H4>
263 <P>The primary anonymous ftp site for PostgreSQL is <A href=
264 "ftp://ftp.PostgreSQL.org/pub">ftp://ftp.PostgreSQL.org/pub</A>.
265 For mirror sites, see our main web site.</P>
267 <H4><A name="1.6">1.6</A>) Where can I get support?</H4>
269 <P>The main mailing list is: <A href=
270 "mailto:pgsql-general@PostgreSQL.org">pgsql-general@PostgreSQL.org</A>.
271 It is available for discussion of matters pertaining to PostgreSQL.
272 To subscribe, send mail with the following lines in the body (not
273 the subject line):</P>
280 "mailto:pgsql-general-request@PostgreSQL.org">pgsql-general-request@PostgreSQL.org</A>.</P>
282 <P>There is also a digest list available. To subscribe to this
283 list, send email to: <A href=
284 "mailto:pgsql-general-digest-request@PostgreSQL.org">pgsql-general-digest-request@PostgreSQL.org</A>
291 Digests are sent out to members of this list whenever the main list
292 has received around 30k of messages.
294 <P>The bugs mailing list is available. To subscribe to this list,
295 send email to <A href=
296 "mailto:pgsql-bugs-request@PostgreSQL.org">pgsql-bugs-request@PostgreSQL.org</A>
303 There is also a developers discussion mailing list available. To
304 subscribe to this list, send email to <A href=
305 "mailto:pgsql-hackers-request@PostgreSQL.org">pgsql-hackers-request@PostgreSQL.org</A>
312 <P>Additional mailing lists and information about PostgreSQL can be
313 found via the PostgreSQL WWW home page at:</P>
316 <A href="http://www.PostgreSQL.org">http://www.PostgreSQL.org</A>
319 <P>There is also an IRC channel on EFNet and OpenProjects,
320 channel <I>#PostgreSQL</I>. I use the Unix command <CODE>irc -c
321 '#PostgreSQL' "$USER" irc.phoenix.net.</CODE></P>
323 <P>A list of commercial support companies is available at <A href=
324 "http://techdocs.postgresql.org/companies.php">http://techdocs.postgresql.org/companies.php</A>.</P>
326 <H4><A name="1.7">1.7</A>) What is the latest release?</H4>
328 <P>The latest release of PostgreSQL is version 7.3.4.</P>
330 <P>We plan to have major releases every six to eight months.</P>
332 <H4><A name="1.8">1.8</A>) What documentation is available?</H4>
334 <P>Several manuals, manual pages, and some small test examples are
335 included in the distribution. See the <I>/doc</I> directory. You
336 can also browse the manuals online at <A href=
337 "http://www.PostgreSQL.org/docs">http://www.PostgreSQL.org/docs</A>.</P>
339 <P>There are two PostgreSQL books available online at <A href=
340 "http://www.PostgreSQL.org/docs/awbook.html">http://www.PostgreSQL.org/docs/awbook.html</A>
342 "http://www.commandprompt.com/ppbook/">http://www.commandprompt.com/ppbook/</A>.
343 There is a list of PostgreSQL books available for purchase at <A
345 "http://techdocs.postgresql.org/techdocs/bookreviews.php">http://techdocs.PostgreSQL.org/techdocs/bookreviews.php</A>.
346 There is also a collection of PostgreSQL technical articles at <A
348 "http://techdocs.PostgreSQL.org/">http://techdocs.PostgreSQL.org/</A>.</P>
350 <P><I>psql</I> has some nice \d commands to show information about
351 types, operators, functions, aggregates, etc.</P>
353 <P>Our web site contains even more documentation.</P>
355 <H4><A name="1.9">1.9</A>) How do I find out about known bugs or
356 missing features?</H4>
358 <P>PostgreSQL supports an extended subset of <SMALL>SQL</SMALL>-92.
359 See our <A href="http://developer.PostgreSQL.org/todo.php">TODO</A>
360 list for known bugs, missing features, and future plans.</P>
362 <H4><A name="1.10">1.10</A>) How can I learn
363 <SMALL>SQL</SMALL>?</H4>
365 <P>The PostgreSQL book at <A href=
366 "http://www.PostgreSQL.org/docs/awbook.html">http://www.PostgreSQL.org/docs/awbook.html</A>
367 teaches <SMALL>SQL</SMALL>. There is another PostgreSQL book at <A
369 "http://www.commandprompt.com/ppbook/">http://www.commandprompt.com/ppbook.</A>
370 There is a nice tutorial at <A href=
371 "http://www.intermedia.net/support/sql/sqltut.shtm">http://www.intermedia.net/support/sql/sqltut.shtm,</A>
373 "http://ourworld.compuserve.com/homepages/graeme_birchall/HTM_COOK.HTM">
374 http://ourworld.compuserve.com/homepages/graeme_birchall/HTM_COOK.HTM,</A>
376 "http://sqlcourse.com/">http://sqlcourse.com.</A></P>
378 <P>Another one is "Teach Yourself SQL in 21 Days, Second Edition"
380 "http://members.tripod.com/er4ebus/sql/index.htm">http://members.tripod.com/er4ebus/sql/index.htm</A></P>
382 <P>Many of our users like <I>The Practical SQL Handbook</I>,
383 Bowman, Judith S., et al., Addison-Wesley. Others like <I>The
384 Complete Reference SQL</I>, Groff et al., McGraw-Hill.</P>
386 <H4><A name="1.11">1.11</A>) Is PostgreSQL Y2K compliant?</H4>
388 <P>Yes, we easily handle dates past the year 2000 AD, and before
391 <H4><A name="1.12">1.12</A>) How do I join the development
394 <P>First, download the latest source and read the PostgreSQL
395 Developers documentation on our web site, or in the distribution.
396 Second, subscribe to the <I>pgsql-hackers</I> and
397 <I>pgsql-patches</I> mailing lists. Third, submit high quality
398 patches to pgsql-patches.</P>
400 <P>There are about a dozen people who have commit privileges to the
401 PostgreSQL <SMALL>CVS</SMALL> archive. They each have submitted so
402 many high-quality patches that it was impossible for the existing
403 committers to keep up, and we had confidence that patches they
404 committed were of high quality.</P>
406 <H4><A name="1.13">1.13</A>) How do I submit a bug report?</H4>
408 <P>Please visit the PostgreSQL BugTool page at <A href=
409 "http://www.PostgreSQL.org/bugs/bugs.php">http://www.PostgreSQL.org/bugs/bugs.php</A>,
410 which gives guidelines and directions on how to submit a
413 <P>Also check out our ftp site <A href=
414 "ftp://ftp.PostgreSQL.org/pub">ftp://ftp.PostgreSQL.org/pub</A> to
415 see if there is a more recent PostgreSQL version or patches.</P>
417 <H4><A name="1.14">1.14</A>) How does PostgreSQL compare to other
418 <SMALL>DBMS</SMALL>s?</H4>
420 <P>There are several ways of measuring software: features,
421 performance, reliability, support, and price.</P>
424 <DT><B>Features</B></DT>
426 <DD>PostgreSQL has most features present in large commercial
427 <SMALL>DBMS</SMALL>s, like transactions, subselects, triggers,
428 views, foreign key referential integrity, and sophisticated
429 locking. We have some features they do not have, like
430 user-defined types, inheritance, rules, and multi-version
431 concurrency control to reduce lock contention.<BR>
435 <DT><B>Performance</B></DT>
437 <DD>PostgreSQL has performance similar to other commercial and
438 open source databases. it is faster for some things, slower for
439 others. In comparison to MySQL or leaner database systems, we are
440 faster for multiple users, complex queries, and a read/write query
441 load. MySQL is faster for SELECT queries done by a few users. Of
442 course, MySQL does not have any of the features mentioned in the
443 <I>Features</I> section above. We are built for reliability and
444 features, though we continue to improve performance in every
445 release. There is an interesting Web page comparing PostgreSQL to
446 MySQL at <A href="http://openacs.org/philosophy/why-not-mysql.html">
447 http://openacs.org/philosophy/why-not-mysql.html</A><BR>
452 <DT><B>Reliability</B></DT>
454 <DD>We realize that a <SMALL>DBMS</SMALL> must be reliable, or it
455 is worthless. We strive to release well-tested, stable code that
456 has a minimum of bugs. Each release has at least one month of
457 beta testing, and our release history shows that we can provide
458 stable, solid releases that are ready for production use. We
459 believe we compare favorably to other database software in this
464 <DT><B>Support</B></DT>
466 <DD>Our mailing lists provide contact with a large group of developers
467 and users to help resolve any problems encountered. While we cannot
468 guarantee a fix, commercial <SMALL>DBMS</SMALL>s do not always
469 supply a fix either. Direct access to developers, the user
470 community, manuals, and the source code often make PostgreSQL
471 support superior to other <SMALL>DBMS</SMALL>s. There is
472 commercial per-incident support available for those who need it.
473 (See <A href="#1.6">FAQ section 1.6</A>.)<BR>
477 <DT><B>Price</B></DT>
479 <DD>We are free for all use, both commercial and non-commercial.
480 You can add our code to your product with no limitations, except
481 those outlined in our BSD-style license stated above.<BR>
486 <H4><A name="1.15">1.15</A>) How can I financially assist
489 <P>PostgreSQL has had a first-class infrastructure since we started
490 in 1996. This is all thanks to Marc Fournier, who has created
491 and managed this infrastructure over the years.</P>
493 <P>Quality infrastructure is very important to an open-source
494 project. It prevents disruptions that can greatly delay forward
495 movement of the project.</P>
497 <P>Of course, this infrastructure is not cheap. There are a variety
498 of monthly and one-time expenses that are required to keep it
499 going. If you or your company has money it can donate to help fund
500 this effort, please go to <A href="http://store.pgsql.com/shopping/">http://store.pgsql.com/shopping/</A>
501 and make a donation.</P>
503 <P>Although the web page mentions PostgreSQL, Inc, the
504 "contributions" item is solely to support the PostgreSQL project
505 and does not fund any specific company. If you prefer, you can also
506 send a check to the contact address.</P>
509 <P>Also, if you have a success story about PostgreSQL, please submit
510 it to our advocacy site at <a href="http://advocacy.postgresql.org">
511 http://advocacy.postgresql.org</a>.</P>
514 <H2 align="center">User Client Questions</H2>
516 <H4><A name="2.1">2.1</A>) Are there <SMALL>ODBC</SMALL> drivers
519 <P>There are two <SMALL>ODBC</SMALL> drivers available, PsqlODBC
520 and OpenLink <SMALL>ODBC</SMALL>.</P>
522 <P>You can download PsqlODBC from <A href=
523 "http://gborg.postgresql.org/project/psqlodbc/projdisplay.php">
524 http://gborg.postgresql.org/project/psqlodbc/projdisplay.php</A>.</P>
526 <P>OpenLink <SMALL>ODBC</SMALL> can be gotten from <A href=
527 "http://www.openlinksw.com/">http://www.openlinksw.com</A>. It
528 works with their standard <SMALL>ODBC</SMALL> client software so
529 you'll have PostgreSQL <SMALL>ODBC</SMALL> available on every
530 client platform they support (Win, Mac, Unix, VMS).</P>
532 <P>They will probably be selling this product to people who need
533 commercial-quality support, but a freeware version will always be
534 available. Please send questions to <A href=
535 "mailto:postgres95@openlink.co.uk">postgres95@openlink.co.uk</A>.</P>
537 <H4><A name="2.2">2.2</A>) What tools are available for using
538 PostgreSQL with Web pages?</H4>
540 <P>A nice introduction to Database-backed Web pages can be seen at:
541 <A href="http://www.webreview.com">http://www.webreview.com</A></P>
543 <P>For Web integration, PHP is an excellent interface. It is at <A
544 href="http://www.php.net">http://www.php.net</A>.</P>
546 <P>For complex cases, many use the Perl interface and CGI.pm or mod_perl.</P>
548 <H4><A name="2.3">2.3</A>) Does PostgreSQL have a graphical user
551 <P>Yes, there are several graphical interfaces to PostgreSQL available.
552 These include PgAccess <a href="http://www.pgaccess.org">
553 http://www.pgaccess.org</a>), PgAdmin II (<a
554 href="http://www.pgadmin.org">http://www.pgadmin.org</a>,
555 Win32-only), RHDB Admin (<a
556 href="http://sources.redhat.com/rhdb/">http://sources.redhat.com/rhdb/
557 </a>) and Rekall (<a href="http://www.thekompany.com/products/rekall/">
558 http://www.thekompany.com/products/rekall/</a>, proprietary). There is
559 also PHPPgAdmin (<a href="http://phppgadmin.sourceforge.net/">
560 http://phppgadmin.sourceforge.net/ </a>), a web-based interface to
563 <P>See <a href="http://techdocs.postgresql.org/guides/GUITools">http://techdocs.postgresql.org/guides/GUITools</a> for a more detailed list.</P>
565 <H4><A name="2.4">2.4</A>) What languages are able to communicate with
568 <P>Most popular programming languages contain an interface to
569 PostgreSQL. Check your programming language's list of extension
572 <P>The following interfaces are included in the PostgreSQL
578 <LI>Embedded C (ecpg)</LI>
582 <LI>Python (PyGreSQL)</LI>
584 <LI>TCL (libpgtcl)</LI>
587 <P>Additional interfaces are available at
588 <a href="http://gborg.postgresql.org">http://gborg.postgresql.org</A>
589 in the <I>Drivers/Interfaces</I> section.
593 <H2 align="center">Administrative Questions</H2>
595 <H4><A name="3.1">3.1</A>) How do I install PostgreSQL somewhere
596 other than <I>/usr/local/pgsql</I>?</H4>
598 <P>Specify the <I>--prefix</I> option when running
599 <I>configure</I>.</P>
601 <H4><A name="3.2">3.2</A>) When I start <I>postmaster</I>, I get a
602 <I>Bad System Call</I> or core dumped message. Why?</H4>
604 <P>It could be a variety of problems, but first check to see that
605 you have System V extensions installed in your kernel. PostgreSQL
606 requires kernel support for shared memory and semaphores.</P>
608 <H4><A name="3.3">3.3</A>) When I try to start <I>postmaster</I>, I
609 get <I>IpcMemoryCreate</I> errors. Why?</H4>
611 <P>You either do not have shared memory configured properly in your
612 kernel or you need to enlarge the shared memory available in the
613 kernel. The exact amount you need depends on your architecture and
614 how many buffers and backend processes you configure for
615 <I>postmaster</I>. For most systems, with default numbers of
616 buffers and processes, you need a minimum of ~1 MB. See the <A
618 "http://www.PostgreSQL.org/docs/view.php?version=current&idoc=1&file=kernel-resources.html">PostgreSQL
619 Administrator's Guide</A> for more detailed information about
620 shared memory and semaphores.</P>
622 <H4><A name="3.4">3.4</A>) When I try to start <I>postmaster</I>, I
623 get <I>IpcSemaphoreCreate</I> errors. Why?</H4>
625 <P>If the error message is <I>IpcSemaphoreCreate: semget failed (No
626 space left on device)</I> then your kernel is not configured with
627 enough semaphores. Postgres needs one semaphore per potential
628 backend process. A temporary solution is to start <I>postmaster</I>
629 with a smaller limit on the number of backend processes. Use
630 <I>-N</I> with a parameter less than the default of 32. A more
631 permanent solution is to increase your kernel's
632 <SMALL>SEMMNS</SMALL> and <SMALL>SEMMNI</SMALL> parameters.</P>
634 <P>Inoperative semaphores can also cause crashes during heavy
637 <P>If the error message is something else, you might not have
638 semaphore support configured in your kernel at all. See the
639 PostgreSQL Administrator's Guide for more detailed information
640 about shared memory and semaphores.</P>
642 <H4><A name="3.5">3.5</A>) How do I control connections from other
645 <P>By default, PostgreSQL only allows connections from the local
646 machine using Unix domain sockets. Other machines will not be able
647 to connect unless you add the <I>-i</I> flag to <I>postmaster</I>,
648 <B>and</B> enable host-based authentication by modifying the file
649 <I>$PGDATA/pg_hba.conf</I> accordingly. This will allow TCP/IP
652 <H4><A name="3.6">3.6</A>) How do I tune the database engine for
653 better performance?</H4>
655 <P>Certainly, indexes can speed up queries. The
656 <SMALL>EXPLAIN</SMALL> command allows you to see how PostgreSQL is
657 interpreting your query, and which indexes are being used.</P>
659 <P>If you are doing many <SMALL>INSERTs</SMALL>, consider doing
660 them in a large batch using the <SMALL>COPY</SMALL> command. This
661 is much faster than individual <SMALL>INSERTS</SMALL>. Second,
662 statements not in a <SMALL>BEGIN WORK/COMMIT</SMALL> transaction
663 block are considered to be in their own transaction. Consider
664 performing several statements in a single transaction block. This
665 reduces the transaction overhead. Also, consider dropping and
666 recreating indexes when making large data changes.</P>
668 <P>There are several tuning options. You can disable <I>fsync()</I>
669 by starting <I>postmaster</I> with a <I>-o -F</I> option. This will
670 prevent <I>fsync()</I>s from flushing to disk after every
673 <P>You can also use the <I>postmaster</I> <I>-B</I> option to
674 increase the number of shared memory buffers used by the backend
675 processes. If you make this parameter too high, the
676 <I>postmaster</I> may not start because you have exceeded your
677 kernel's limit on shared memory space. Each buffer is 8K and the
678 default is 64 buffers.</P>
680 <P>You can also use the backend <I>-S</I> option to increase the
681 maximum amount of memory used by the backend process for temporary
682 sorts. The <I>-S</I> value is measured in kilobytes, and the
683 default is 512 (i.e. 512K).</P>
685 <P>You can also use the <SMALL>CLUSTER</SMALL> command to group
686 data in tables to match an index. See the <SMALL>CLUSTER</SMALL>
687 manual page for more details.</P>
689 <H4><A name="3.7">3.7</A>) What debugging features are
692 <P>PostgreSQL has several features that report status information
693 that can be valuable for debugging purposes.</P>
695 <P>First, by running <I>configure</I> with the --enable-cassert
696 option, many <I>assert()</I>s monitor the progress of the backend
697 and halt the program when something unexpected occurs.</P>
699 <P>Both <I>postmaster</I> and <I>postgres</I> have several debug
700 options available. First, whenever you start <I>postmaster</I>,
701 make sure you send the standard output and error to a log file,
705 ./bin/postmaster >server.log 2>&1 &
708 <P>This will put a server.log file in the top-level PostgreSQL
709 directory. This file contains useful information about problems or
710 errors encountered by the server. <I>Postmaster</I> has a <I>-d</I>
711 option that allows even more detailed information to be reported.
712 The <I>-d</I> option takes a number that specifies the debug level.
713 Be warned that high debug level values generate large log
716 <P>If <I>postmaster</I> is not running, you can actually run the
717 <I>postgres</I> backend from the command line, and type your
718 <SMALL>SQL</SMALL> statement directly. This is recommended
719 <B>only</B> for debugging purposes. Note that a newline terminates
720 the query, not a semicolon. If you have compiled with debugging
721 symbols, you can use a debugger to see what is happening. Because
722 the backend was not started from <I>postmaster</I>, it is not
723 running in an identical environment and locking/backend interaction
724 problems may not be duplicated.</P>
726 <P>If <I>postmaster</I> is running, start <I>psql</I> in one
727 window, then find the <SMALL>PID</SMALL> of the <I>postgres</I>
728 process used by <I>psql</I>. Use a debugger to attach to the
729 <I>postgres</I> <SMALL>PID</SMALL>. You can set breakpoints in the
730 debugger and issue queries from <I>psql</I>. If you are debugging
731 <I>postgres</I> startup, you can set PGOPTIONS="-W n", then start
732 <I>psql</I>. This will cause startup to delay for <I>n</I> seconds
733 so you can attach to the process with the debugger, set any
734 breakpoints, and continue through the startup sequence.</P>
736 <P>The <I>postgres</I> program has <I>-s, -A</I>, and <I>-t</I>
737 options that can be very useful for debugging and performance
740 <P>You can also compile with profiling to see what functions are
741 taking execution time. The backend profile files will be deposited
742 in the <I>pgsql/data/base/dbname</I> directory. The client profile
743 file will be put in the client's current directory. Linux requires
744 a compile with <I>-DLINUX_PROFILE</I> for proper profiling.</P>
746 <H4><A name="3.8">3.8</A>) Why do I get <I>"Sorry, too many
747 clients"</I> when trying to connect?</H4>
749 <P>You need to increase <I>postmaster</I>'s limit on how many
750 concurrent backend processes it can start.</P>
752 <P>The default limit is 32 processes. You can increase it by
753 restarting <I>postmaster</I> with a suitable <I>-N</I> value or
754 modifying <I>postgresql.conf</I>.</P>
756 <P>Note that if you make <I>-N</I> larger than 32, you must also
757 increase <I>-B</I> beyond its default of 64; <I>-B</I> must be at
758 least twice <I>-N</I>, and probably should be more than that for
759 best performance. For large numbers of backend processes, you are
760 also likely to find that you need to increase various Unix kernel
761 configuration parameters. Things to check include the maximum size
762 of shared memory blocks, <SMALL>SHMMAX;</SMALL> the maximum number
763 of semaphores, <SMALL>SEMMNS</SMALL> and <SMALL>SEMMNI;</SMALL> the
764 maximum number of processes, <SMALL>NPROC;</SMALL> the maximum
765 number of processes per user, <SMALL>MAXUPRC;</SMALL> and the
766 maximum number of open files, <SMALL>NFILE</SMALL> and
767 <SMALL>NINODE</SMALL>. The reason that PostgreSQL has a limit on
768 the number of allowed backend processes is so your system won't run
769 out of resources.</P>
771 <H4><A name="3.9">3.9</A>) What is in the <I>pgsql_tmp</I> directory?</H4>
773 <P>This directory contains temporary files generated by the query
774 executor. For example, if a sort needs to be done to satisfy an
775 <SMALL>ORDER BY</SMALL> and the sort requires more space than the
776 backend's <I>-S</I> parameter allows, then temporary files are created
777 here to hold the extra data.</P>
779 <P>The temporary files are usually deleted automatically, but might
780 remain if a backend crashes during a sort. A stop and restart of the
781 <I>postmaster</I> will remove files from those directories.</P>
783 <H4><A name="3.10">3.10</A>) Why do I need to do a dump and restore
784 to upgrade between major PostgreSQL releases?</H4>
786 <P>The PostgreSQL team makes only small changes between minor releases,
787 so upgrading from 7.2 to 7.2.1 does not require a dump and restore.
788 However, major releases (e.g. from 7.2 to 7.3) often change the internal
789 format of system tables and data files. These changes are often complex,
790 so we don't maintain backward compatability for data files. A dump outputs
791 data in a generic format that can then be loaded in using the new internal
794 <P>In releases where the on-disk format does not change, the
795 <I>pg_upgrade</I> script can be used to upgrade without a dump/restore.
796 The release notes mention whether <I>pg_upgrade</I> is available for the
801 <H2 align="center">Operational Questions</H2>
803 <H4><A name="4.1">4.1</A>) What is the difference between binary
804 cursors and normal cursors?</H4>
806 <P>See the <SMALL>DECLARE</SMALL> manual page for a
809 <H4><A name="4.2">4.2</A>) How do I <SMALL>SELECT</SMALL> only the
810 first few rows of a query? A random row?</H4>
812 <P>See the <SMALL>FETCH</SMALL> manual page, or use
813 <SMALL>SELECT</SMALL> ... <SMALL>LIMIT</SMALL>....</P>
815 <P>The entire query may have to be evaluated, even if you only want
816 the first few rows. Consider using a query that has an <SMALL>ORDER
817 BY</SMALL>. If there is an index that matches the <SMALL>ORDER
818 BY</SMALL>, PostgreSQL may be able to evaluate only the first few
819 records requested, or the entire query may have to be evaluated
820 until the desired rows have been generated.</P>
822 <P>To <SMALL>SELECT</SMALL> a random row, use:
830 <H4><A name="4.3">4.3</A>) How do I get a list of tables or other
831 things I can see in <I>psql</I>?</H4>
833 <P>You can read the source code for <I>psql</I> in file
834 <I>pgsql/src/bin/psql/describe.c</I>. It contains
835 <SMALL>SQL</SMALL> commands that generate the output for psql's
836 backslash commands. You can also start <I>psql</I> with the
837 <I>-E</I> option so it will print out the queries it uses to
838 execute the commands you give.</P>
840 <H4><A name="4.4">4.4</A>) How do you remove a column from a
841 table, or change its data type?</H4>
843 <P><SMALL>DROP COLUMN</SMALL> functionality was added in release 7.3 with
844 <SMALL>ALTER TABLE DROP COLUMN</SMALL>. In earlier versions,
848 LOCK TABLE old_table;
849 SELECT ... -- select all columns but the one you want to remove
852 DROP TABLE old_table;
853 ALTER TABLE new_table RENAME TO old_table;
857 <P>To change the data type of a column, do this:</P>
860 ALTER TABLE tab ADD COLUMN new_col <i>new_data_type</i>;
861 UPDATE tab SET new_col = CAST(old_col AS <i>new_data_type</i>);
862 ALTER TABLE tab DROP COLUMN old_col;
865 <P>You might then want to do <I>VACUUM FULL tab</I> to reclaim the
866 disk space used by the expired rows.</P>
868 <H4><A name="4.5">4.5</A>) What is the maximum size for a row, a
869 table, and a database?</H4>
871 <P>These are the limits:</P>
873 Maximum size for a database? unlimited (4 TB databases exist)
874 Maximum size for a table? 32 TB
875 Maximum size for a row? 1.6TB
876 Maximum size for a field? 1 GB
877 Maximum number of rows in a table? unlimited
878 Maximum number of columns in a table? 250-1600 depending on column types
879 Maximum number of indexes on a table? unlimited
882 Of course, these are not actually unlimited, but limited to
883 available disk space and memory/swap space. Performance may suffer
884 when these values get unusually large.
886 <P>The maximum table size of 32 TB does not require large file
887 support from the operating system. Large tables are stored as
888 multiple 1 GB files so file system size limits are not
891 <P>The maximum table size and maximum number of columns can be
892 quadrupled by increasing the default block size to 32k.</P>
894 <H4><A name="4.6">4.6</A>) How much database disk space is required
895 to store data from a typical text file?</H4>
897 <P>A PostgreSQL database may require up to five times the disk
898 space to store data from a text file.</P>
900 <P>As an example, consider a file of 100,000 lines with an integer
901 and text description on each line. Suppose the text string
902 avergages twenty bytes in length. The flat file would be 2.8 MB.
903 The size of the PostgreSQL database file containing this data can
904 be estimated as 6.4 MB:</P>
906 36 bytes: each row header (approximate)
907 24 bytes: one int field and one text field
908 + 4 bytes: pointer on page to tuple
909 ----------------------------------------
912 The data page size in PostgreSQL is 8192 bytes (8 KB), so:
915 ------------------- = 128 rows per database page (rounded down)
919 -------------------- = 782 database pages (rounded up)
922 782 database pages * 8192 bytes per page = 6,406,144 bytes (6.4 MB)
925 <P>Indexes do not require as much overhead, but do contain the data
926 that is being indexed, so they can be large also.</P>
928 <P><SMALL>NULL</SMALL>s are stored in bitmaps, so they
929 use very little space.</P>
931 <H4><A name="4.7">4.7</A>) How do I find out what tables, indexes,
932 databases, and users are defined?</H4>
934 <P><I>psql</I> has a variety of backslash commands to show such
935 information. Use \? to see them. There are also system tables
936 beginning with <I>pg_</I> that describe these too. Also, <I>psql
937 -l</I> will list all databases.</P>
939 <P>Also try the file <I>pgsql/src/tutorial/syscat.source</I>. It
940 illustrates many of the <SMALL>SELECT</SMALL>s needed to get
941 information from the database system tables.</P>
943 <H4><A name="4.8">4.8</A>) My queries are slow or don't make use of
944 the indexes. Why?</H4>
945 Indexes are not automatically used by every query. Indexes are only
946 used if the table is larger than a minimum size, and the query
947 selects only a small percentage of the rows in the table. This is
948 because the random disk access caused by an index scan can be
949 slower than a straight read through the table, or sequential scan.
951 <P>To determine if an index should be used, PostgreSQL must have
952 statistics about the table. These statistics are collected using
953 <SMALL>VACUUM ANALYZE</SMALL>, or simply <SMALL>ANALYZE</SMALL>.
954 Using statistics, the optimizer knows how many rows are in the
955 table, and can better determine if indexes should be used.
956 Statistics are also valuable in determining optimal join order and
957 join methods. Statistics collection should be performed
958 periodically as the contents of the table change.</P>
960 <P>Indexes are normally not used for <SMALL>ORDER BY</SMALL> or to
961 perform joins. A sequential scan followed by an explicit sort is
962 usually faster than an index scan of a large table.</P>
963 However, <SMALL>LIMIT</SMALL> combined with <SMALL>ORDER BY</SMALL>
964 often will use an index because only a small portion of the table
965 is returned. In fact, though MAX() and MIN() don't use indexes,
966 it is possible to retrieve such values using an index with ORDER BY
971 ORDER BY col [ DESC ]
975 <P>If you believe the optimizer is incorrect in choosing a
976 sequential scan, use <CODE>SET enable_seqscan TO 'off'</CODE> and
977 run tests to see if an index scan is indeed faster.</P>
979 <P>When using wild-card operators such as <SMALL>LIKE</SMALL> or
980 <I>~</I>, indexes can only be used in certain circumstances:</P>
982 <LI>The beginning of the search string must be anchored to the start
985 <LI><SMALL>LIKE</SMALL> patterns must not start with <I>%</I>.</LI>
986 <LI><I>~</I> (regular expression) patterns must start with
989 <LI>The search string can not start with a character class,
991 <LI>Case-insensitive searches such as <SMALL>ILIKE</SMALL> and
992 <I>~*</I> do not utilise indexes. Instead, use functional
993 indexes, which are described in section <a href="#4.12">4.12</a>.</LI>
994 <LI>The default <I>C</I> locale must be used during
999 <H4><A name="4.9">4.9</A>) How do I see how the query optimizer is
1000 evaluating my query?</H4>
1002 <P>See the <SMALL>EXPLAIN</SMALL> manual page.</P>
1004 <H4><A name="4.10">4.10</A>) What is an R-tree index?</H4>
1006 <P>An R-tree index is used for indexing spatial data. A hash index
1007 can't handle range searches. A B-tree index only handles range
1008 searches in a single dimension. R-trees can handle
1009 multi-dimensional data. For example, if an R-tree index can be
1010 built on an attribute of type <I>point</I>, the system can more
1011 efficiently answer queries such as "select all points within a
1012 bounding rectangle."</P>
1014 <P>The canonical paper that describes the original R-tree design
1017 <P>Guttman, A. "R-trees: A Dynamic Index Structure for Spatial
1018 Searching." Proceedings of the 1984 ACM SIGMOD Int'l Conf on Mgmt
1021 <P>You can also find this paper in Stonebraker's "Readings in
1022 Database Systems".</P>
1024 <P>Built-in R-trees can handle polygons and boxes. In theory,
1025 R-trees can be extended to handle higher number of dimensions. In
1026 practice, extending R-trees requires a bit of work and we don't
1027 currently have any documentation on how to do it.</P>
1029 <H4><A name="4.11">4.11</A>) What is the Genetic Query
1032 <P>The <SMALL>GEQO</SMALL> module speeds query optimization when
1033 joining many tables by means of a Genetic Algorithm (GA). It allows
1034 the handling of large join queries through nonexhaustive
1037 <H4><A name="4.12">4.12</A>) How do I perform regular expression
1038 searches and case-insensitive regular expression searches? How do I
1039 use an index for case-insensitive searches?</H4>
1041 <P>The <I>~</I> operator does regular expression matching, and
1042 <I>~*</I> does case-insensitive regular expression matching. The
1043 case-insensitive variant of <SMALL>LIKE</SMALL> is called
1044 <SMALL>ILIKE</SMALL>.</P>
1046 <P>Case-insensitive equality comparisons are normally expressed
1051 WHERE lower(col) = 'abc';
1054 This will not use an standard index. However, if you create a
1055 functional index, it will be used:
1057 CREATE INDEX tabindex ON tab (lower(col));
1060 <H4><A name="4.13">4.13</A>) In a query, how do I detect if a field
1061 is <SMALL>NULL</SMALL>?</H4>
1063 <P>You test the column with <SMALL>IS NULL</SMALL> and <SMALL>IS
1064 NOT NULL</SMALL>.</P>
1066 <H4><A name="4.14">4.14</A>) What is the difference between the
1067 various character types?</H4>
1069 Type Internal Name Notes
1070 --------------------------------------------------
1071 VARCHAR(n) varchar size specifies maximum length, no padding
1072 CHAR(n) bpchar blank padded to the specified fixed length
1073 TEXT text no specific upper limit on length
1074 BYTEA bytea variable-length byte array (null-byte safe)
1075 "char" char one character
1078 <P>You will see the internal name when examining system catalogs
1079 and in some error messages.</P>
1081 <P>The first four types above are "varlena" types (i.e., the first
1082 four bytes on disk are the length, followed by the data). Thus the
1083 actual space used is slightly greater than the declared size.
1084 However, these data types are also subject to compression or being
1085 stored out-of-line by <SMALL>TOAST</SMALL>, so the space on disk
1086 might also be less than expected.</P>
1088 <SMALL>VARCHAR(n)</SMALL> is best when storing variable-length
1089 strings and it limits how long a string can be. <SMALL>TEXT</SMALL>
1090 is for strings of unlimited length, with a maximum of one gigabyte.
1091 <P><SMALL>CHAR(n)</SMALL> is for storing strings that are all the
1092 same length. <SMALL>CHAR(n)</SMALL> pads with blanks to the specified
1093 length, while <SMALL>VARCHAR(n)</SMALL> only stores the characters
1094 supplied. <SMALL>BYTEA</SMALL> is for storing binary data,
1095 particularly values that include <SMALL>NULL</SMALL> bytes. All the
1096 types described here have similar performance characteristics.</P>
1098 <H4><A name="4.15.1">4.15.1</A>) How do I create a
1099 serial/auto-incrementing field?</H4>
1101 <P>PostgreSQL supports a <SMALL>SERIAL</SMALL> data type. It
1102 auto-creates a sequence and index on the column. For example,
1105 CREATE TABLE person (
1111 is automatically translated into this:
1113 CREATE SEQUENCE person_id_seq;
1114 CREATE TABLE person (
1115 id INT4 NOT NULL DEFAULT nextval('person_id_seq'),
1118 CREATE UNIQUE INDEX person_id_key ON person ( id );
1121 See the <I>create_sequence</I> manual page for more information
1122 about sequences. You can also use each row's <I>OID</I> field as a
1123 unique value. However, if you need to dump and reload the database,
1124 you need to use <I>pg_dump</I>'s <I>-o</I> option or <SMALL>COPY
1125 WITH OIDS</SMALL> option to preserve the <SMALL>OID</SMALL>s.
1127 <H4><A name="4.15.2">4.15.2</A>) How do I get the value of a
1128 <SMALL>SERIAL</SMALL> insert?</H4>
1130 <P>One approach is to retrieve the next <SMALL>SERIAL</SMALL> value
1131 from the sequence object with the <I>nextval()</I> function
1132 <I>before</I> inserting and then insert it explicitly. Using the
1133 example table in <A href="#4.15.1">4.15.1</A>, an example in a
1134 pseudo-language would look like this:</P>
1136 new_id = execute("SELECT nextval('person_id_seq')");
1137 execute("INSERT INTO person (id, name) VALUES (new_id, 'Blaise Pascal')");
1140 You would then also have the new value stored in
1141 <CODE>new_id</CODE> for use in other queries (e.g., as a foreign
1142 key to the <CODE>person</CODE> table). Note that the name of the
1143 automatically created <SMALL>SEQUENCE</SMALL> object will be named
1144 <<I>table</I>>_<<I>serialcolumn</I>>_<I>seq</I>, where
1145 <I>table</I> and <I>serialcolumn</I> are the names of your table
1146 and your <SMALL>SERIAL</SMALL> column, respectively.
1148 <P>Alternatively, you could retrieve the assigned
1149 <SMALL>SERIAL</SMALL> value with the <I>currval()</I> function
1150 <I>after</I> it was inserted by default, e.g.,</P>
1152 execute("INSERT INTO person (name) VALUES ('Blaise Pascal')");
1153 new_id = execute("SELECT currval('person_id_seq')");
1156 Finally, you could use the <A href="#4.16"><SMALL>OID</SMALL></A>
1157 returned from the <SMALL>INSERT</SMALL> statement to look up the
1158 default value, though this is probably the least portable approach,
1159 and the oid value will wrap around when it reaches 4 billion.
1160 In Perl, using DBI with Edmund Mergl's DBD::Pg module, the oid
1161 value is made available via <I>$sth->{pg_oid_status}</I> after
1162 <I>$sth->execute()</I>.
1164 <H4><A name="4.15.3">4.15.3</A>) Don't <I>currval()</I> and
1165 <I>nextval()</I> lead to a race condition with other users?</H4>
1167 <P>No. <I>currval()</I> returns the current value assigned by your
1168 backend, not by all users.</P>
1170 <H4><A name="4.15.4">4.15.4</A>) Why aren't my sequence numbers
1171 reused on transaction abort? Why are there gaps in the numbering of
1172 my sequence/SERIAL column?</H4>
1174 <P>To improve concurrency, sequence values are given out to running
1175 transactions as needed and are not locked until the transaction
1176 completes. This causes gaps in numbering from aborted
1179 <H4><A name="4.16">4.16</A>) What is an <SMALL>OID</SMALL>? What is
1180 a <SMALL>TID</SMALL>?</H4>
1182 <P><SMALL>OID</SMALL>s are PostgreSQL's answer to unique row ids.
1183 Every row that is created in PostgreSQL gets a unique
1184 <SMALL>OID</SMALL>. All <SMALL>OID</SMALL>s generated during
1185 <I>initdb</I> are less than 16384 (from
1186 <I>include/access/transam.h</I>). All user-created
1187 <SMALL>OID</SMALL>s are equal to or greater than this. By default,
1188 all these <SMALL>OID</SMALL>s are unique not only within a table or
1189 database, but unique within the entire PostgreSQL installation.</P>
1191 <P>PostgreSQL uses <SMALL>OID</SMALL>s in its internal system
1192 tables to link rows between tables. These <SMALL>OID</SMALL>s can
1193 be used to identify specific user rows and used in joins. It is
1194 recommended you use column type <SMALL>OID</SMALL> to store
1195 <SMALL>OID</SMALL> values. You can create an index on the
1196 <SMALL>OID</SMALL> field for faster access.</P>
1198 <P>O<SMALL>ID</SMALL>s are assigned to all new rows from a central
1199 area that is used by all databases. If you want to change the
1200 <SMALL>OID</SMALL> to something else, or if you want to make a copy
1201 of the table, with the original <SMALL>OID</SMALL>s, there is no
1202 reason you can't do it:</P>
1204 CREATE TABLE new_table(old_oid oid, mycol int);
1205 SELECT old_oid, mycol INTO new FROM old;
1206 COPY new TO '/tmp/pgtable';
1208 COPY new WITH OIDS FROM '/tmp/pgtable';
1211 CREATE TABLE new_table (mycol int);
1212 INSERT INTO new_table (oid, mycol) SELECT oid, mycol FROM old_table;
1214 <P>O<SMALL>ID</SMALL>s are stored as 4-byte integers, and will
1215 overflow at 4 billion. No one has reported this ever happening, and
1216 we plan to have the limit removed before anyone does.</P>
1218 <P>T<SMALL>ID</SMALL>s are used to identify specific physical rows
1219 with block and offset values. T<SMALL>ID</SMALL>s change after rows
1220 are modified or reloaded. They are used by index entries to point
1221 to physical rows.</P>
1223 <H4><A name="4.17">4.17</A>) What is the meaning of some of the
1224 terms used in PostgreSQL?</H4>
1226 <P>Some of the source code and older documentation use terms that
1227 have more common usage. Here are some:</P>
1230 <LI>table, relation, class</LI>
1232 <LI>row, record, tuple</LI>
1234 <LI>column, field, attribute</LI>
1236 <LI>retrieve, select</LI>
1238 <LI>replace, update</LI>
1240 <LI>append, insert</LI>
1242 <LI><SMALL>OID</SMALL>, serial value</LI>
1244 <LI>portal, cursor</LI>
1246 <LI>range variable, table name, table alias</LI>
1249 <P>A list of general database terms can be found at: <A href=
1250 "http://hea-www.harvard.edu/MST/simul/software/docs/pkgs/pgsql/glossary/glossary.html">http://hea-www.harvard.edu/MST/simul/software/docs/pkgs/pgsql/glossary/glossary.html</A></P>
1252 <H4><A name="4.18">4.18</A>) Why do I get the error <I>"ERROR:
1253 Memory exhausted in AllocSetAlloc()"</I>?</H4>
1255 <P>You probably have run out of virtual memory on your system,
1256 or your kernel has a low limit for certain resources. Try this
1257 before starting <I>postmaster</I>:</P>
1263 Depending on your shell, only one of these may succeed, but it will
1264 set your process data segment limit much higher and perhaps allow
1265 the query to complete. This command applies to the current process,
1266 and all subprocesses created after the command is run. If you are
1267 having a problem with the <SMALL>SQL</SMALL> client because the
1268 backend is returning too much data, try it before starting the
1271 <H4><A name="4.19">4.19</A>) How do I tell what PostgreSQL version
1274 <P>From <I>psql</I>, type <CODE>SELECT version();</CODE></P>
1276 <H4><A name="4.20">4.20</A>) Why does my large-object operations
1277 get <I>"invalid large obj descriptor"</I>?</H4>
1279 <P>You need to put <CODE>BEGIN WORK</CODE> and <CODE>COMMIT</CODE>
1280 around any use of a large object handle, that is, surrounding
1281 <CODE>lo_open</CODE> ... <CODE>lo_close.</CODE></P>
1283 <P>Currently PostgreSQL enforces the rule by closing large object
1284 handles at transaction commit. So the first attempt to do anything
1285 with the handle will draw <I>invalid large obj descriptor</I>. So
1286 code that used to work (at least most of the time) will now
1287 generate that error message if you fail to use a transaction.</P>
1289 <P>If you are using a client interface like <SMALL>ODBC</SMALL> you
1290 may need to set <CODE>auto-commit off.</CODE></P>
1292 <H4><A name="4.21">4.21</A>) How do I create a column that will
1293 default to the current time?</H4>
1295 <P>Use <I>CURRENT_TIMESTAMP</I>:</P>
1297 <CODE>CREATE TABLE test (x int, modtime timestamp DEFAULT CURRENT_TIMESTAMP );
1301 <H4><A name="4.22">4.22</A>) Why are my subqueries using
1302 <CODE><SMALL>IN</SMALL></CODE> so slow?</H4>
1304 <P>Currently, we join subqueries to outer queries by sequentially
1305 scanning the result of the subquery for each row of the outer
1306 query. If the subquery returns only a few rows and the outer query
1307 returns many rows, <CODE><SMALL>IN</SMALL></CODE> is fastest. To
1308 speed up other queries, replace <CODE>IN</CODE> with
1309 <CODE>EXISTS</CODE>:</P>
1312 WHERE col IN (SELECT subcol FROM subtab);
1317 WHERE EXISTS (SELECT subcol FROM subtab WHERE subcol = col);
1320 For this to be fast, <CODE>subcol</CODE> should be an indexed column.
1321 This preformance problem will be fixed in 7.4.
1323 <H4><A name="4.23">4.23</A>) How do I perform an outer join?</H4>
1325 <P>PostgreSQL supports outer joins using the SQL standard syntax.
1326 Here are two examples:</P>
1329 FROM t1 LEFT OUTER JOIN t2 ON (t1.col = t2.col);
1334 FROM t1 LEFT OUTER JOIN t2 USING (col);
1337 <P>These identical queries join t1.col to t2.col, and also return
1338 any unjoined rows in t1 (those with no match in t2). A
1339 <SMALL>RIGHT</SMALL> join would add unjoined rows of t2. A
1340 <SMALL>FULL</SMALL> join would return the matched rows plus all
1341 unjoined rows from t1 and t2. The word <SMALL>OUTER</SMALL> is
1342 optional and is assumed in <SMALL>LEFT</SMALL>,
1343 <SMALL>RIGHT</SMALL>, and <SMALL>FULL</SMALL> joins. Ordinary joins
1344 are called <SMALL>INNER</SMALL> joins.</P>
1346 <P>In previous releases, outer joins can be simulated using
1347 <SMALL>UNION</SMALL> and <SMALL>NOT IN</SMALL>. For example, when
1348 joining <I>tab1</I> and <I>tab2</I>, the following query does an
1349 <I>outer</I> join of the two tables:<BR>
1353 SELECT tab1.col1, tab2.col2
1355 WHERE tab1.col1 = tab2.col1
1357 SELECT tab1.col1, NULL
1359 WHERE tab1.col1 NOT IN (SELECT tab2.col1 FROM tab2)
1363 <H4><A name="4.24">4.24</A>) How do I perform queries using
1364 multiple databases?</H4>
1366 <P>There is no way to query a database other than the current one.
1367 Because PostgreSQL loads database-specific system catalogs, it is
1368 uncertain how a cross-database query should even behave.</P>
1370 <P><I>contrib/dblink</I> allows cross-database queries using
1371 function calls. Of course, a client can make simultaneous
1372 connections to different databases and merge the results on the
1375 <H4><A name="4.25">4.25</A>) How do I return multiple rows or
1376 columns from a function?</H4>
1378 <P>In 7.3, you can easily return multiple rows or columns from a
1380 <a href="http://techdocs.postgresql.org/guides/SetReturningFunctions">
1381 http://techdocs.postgresql.org/guides/SetReturningFunctions</a>.
1383 <H4><A name="4.26">4.26</A>) Why can't I reliably create/drop
1384 temporary tables in PL/PgSQL functions?</H4>
1385 <P>PL/PgSQL caches function contents, and an unfortunate side effect
1386 is that if a PL/PgSQL function accesses a temporary table, and that
1387 table is later dropped and recreated, and the function called
1388 again, the function will fail because the cached function contents
1389 still point to the old temporary table. The solution is to use
1390 <SMALL>EXECUTE</SMALL> for temporary table access in PL/PgSQL. This
1391 will cause the query to be reparsed every time.</P>
1393 <H4><A name="4.27">4.27</A>) What replication options are available?
1395 <P>There are several master/slave replication options available.
1396 These allow only the master to make database changes and the slave
1397 can only do database reads. The bottom of <a
1398 href="http://gborg.PostgreSQL.org/genpage?replication_research">
1399 http://gborg.PostgreSQL.org/genpage?replication_research</a> lists
1400 them. A multi-master replication solution is being worked on at <a
1401 href="http://gborg.PostgreSQL.org/project/pgreplication/projdisplay.php">http://gborg.PostgreSQL.org/project/pgreplication/projdisplay.php</a>.</P>
1403 <H4><A name="4.28">4.28</A>) What encryption options are available?
1406 <LI><I>contrib/pgcrypto</I> contains many encryption functions for
1407 use in <SMALL>SQL</SMALL> queries.</LI>
1408 <LI>To encrypt transmission from the client to the server, the server
1409 must have the <I>ssl</I> option set to <I>true</I> in <I>postgresql.conf,
1410 </I> and an applicable <I>host</I> or <I>hostssl</I> record must exist in
1411 <I>pg_hba.conf</I>, and the client <I>sslmode</I> must not be
1412 <I>disable.</I> (Note that it is also possible to use a third-party
1413 encrypted transport, such as stunnel or ssh, rather than PostgreSQL's
1414 native SSL connections.)
1415 <LI>Database user passwords are automatically encrypted when stored in
1416 version 7.3. In previous versions, you must enable the option
1417 <I>PASSWORD_ENCRYPTION</I> in <I>postgresql.conf</I>.</LI>
1418 <LI>The server can run using an encrypted file system.</LI>
1423 <H2 align="center">Extending PostgreSQL</H2>
1425 <H4><A name="5.1">5.1</A>) I wrote a user-defined function. When I
1426 run it in <I>psql</I>, why does it dump core?</H4>
1428 <P>The problem could be a number of things. Try testing your
1429 user-defined function in a stand-alone test program first.</P>
1431 <H4><A name="5.2">5.2</A>) How can I contribute some nifty new
1432 types and functions to PostgreSQL?</H4>
1434 <P>Send your extensions to the <I>pgsql-hackers</I> mailing list,
1435 and they will eventually end up in the <I>contrib/</I>
1438 <H4><A name="5.3">5.3</A>) How do I write a C function to return a
1441 <P>In versions of PostgreSQL beginning with 7.3, table-returning
1442 functions are fully supported in C, PL/PgSQL, and SQL. See the
1443 Programmer's Guide for more information. An example of a
1444 table-returning function defined in C can be found in
1445 <I>contrib/tablefunc</I>.</P>
1447 <H4><A name="5.4">5.4</A>) I have changed a source file. Why does
1448 the recompile not see the change?</H4>
1450 <P>The <I>Makefiles</I> do not have the proper dependencies for
1451 include files. You have to do a <I>make clean</I> and then another
1452 <I>make</I>. If you are using <SMALL>GCC</SMALL> you can use the
1453 <I>--enable-depend</I> option of <I>configure</I> to have the
1454 compiler compute the dependencies automatically.</P>