The code to unlink dropped relations in FinishPreparedTransaction() was
[PostgreSQL.git] / doc / src / sgml / ref / pg_restore.sgml
blob35089e73e0c6dcd68994597d0ae07ce7cc41998b
1 <!-- $PostgreSQL$ -->
3 <refentry id="APP-PGRESTORE">
4 <refmeta>
5 <refentrytitle>pg_restore</refentrytitle>
6 <manvolnum>1</manvolnum>
7 <refmiscinfo>Application</refmiscinfo>
8 </refmeta>
10 <refnamediv>
11 <refname>pg_restore</refname>
13 <refpurpose>
14 restore a <productname>PostgreSQL</productname> database from an
15 archive file created by <application>pg_dump</application>
16 </refpurpose>
17 </refnamediv>
19 <indexterm zone="app-pgrestore">
20 <primary>pg_restore</primary>
21 </indexterm>
23 <refsynopsisdiv>
24 <cmdsynopsis>
25 <command>pg_restore</command>
26 <arg rep="repeat"><replaceable>option</replaceable></arg>
27 <arg><replaceable>filename</replaceable></arg>
28 </cmdsynopsis>
29 </refsynopsisdiv>
32 <refsect1 id="app-pgrestore-description">
33 <title>Description</title>
35 <para>
36 <application>pg_restore</application> is a utility for restoring a
37 <productname>PostgreSQL</productname> database from an archive
38 created by <xref linkend="app-pgdump"> in one of the non-plain-text
39 formats. It will issue the commands necessary to reconstruct the
40 database to the state it was in at the time it was saved. The
41 archive files also allow <application>pg_restore</application> to
42 be selective about what is restored, or even to reorder the items
43 prior to being restored. The archive files are designed to be
44 portable across architectures.
45 </para>
47 <para>
48 <application>pg_restore</application> can operate in two modes.
49 If a database name is specified, the archive is restored directly into
50 the database. Otherwise, a script containing the SQL
51 commands necessary to rebuild the database is created and written
52 to a file or standard output. The script output is equivalent to
53 the plain text output format of <application>pg_dump</application>.
54 Some of the options controlling the output are therefore analogous to
55 <application>pg_dump</application> options.
56 </para>
58 <para>
59 Obviously, <application>pg_restore</application> cannot restore information
60 that is not present in the archive file. For instance, if the
61 archive was made using the <quote>dump data as
62 <command>INSERT</command> commands</quote> option,
63 <application>pg_restore</application> will not be able to load the data
64 using <command>COPY</command> statements.
65 </para>
66 </refsect1>
68 <refsect1 id="app-pgrestore-options">
69 <title>Options</title>
71 <para>
72 <application>pg_restore</application> accepts the following command
73 line arguments.
75 <variablelist>
76 <varlistentry>
77 <term><replaceable class="parameter">filename</replaceable></term>
78 <listitem>
79 <para>
80 Specifies the location of the archive file to be restored.
81 If not specified, the standard input is used.
82 </para>
83 </listitem>
84 </varlistentry>
86 <varlistentry>
87 <term><option>-a</option></term>
88 <term><option>--data-only</option></term>
89 <listitem>
90 <para>
91 Restore only the data, not the schema (data definitions).
92 </para>
93 </listitem>
94 </varlistentry>
96 <varlistentry>
97 <term><option>-c</option></term>
98 <term><option>--clean</option></term>
99 <listitem>
100 <para>
101 Clean (drop) database objects before recreating them.
102 </para>
103 </listitem>
104 </varlistentry>
106 <varlistentry>
107 <term><option>-C</option></term>
108 <term><option>--create</option></term>
109 <listitem>
110 <para>
111 Create the database before restoring into it. (When this
112 option is used, the database named with <option>-d</option> is
113 used only to issue the initial <command>CREATE DATABASE</>
114 command. All data is restored into the database name that
115 appears in the archive.)
116 </para>
117 </listitem>
118 </varlistentry>
120 <varlistentry>
121 <term><option>-d <replaceable class="parameter">dbname</replaceable></option></term>
122 <term><option>--dbname=<replaceable class="parameter">dbname</replaceable></option></term>
123 <listitem>
124 <para>
125 Connect to database <replaceable
126 class="parameter">dbname</replaceable> and restore directly
127 into the database.
128 </para>
129 </listitem>
130 </varlistentry>
132 <varlistentry>
133 <term><option>-e</option></term>
134 <term><option>--exit-on-error</option></term>
135 <listitem>
136 <para>
137 Exit if an error is encountered while sending SQL commands to
138 the database. The default is to continue and to display a count of
139 errors at the end of the restoration.
140 </para>
141 </listitem>
142 </varlistentry>
144 <varlistentry>
145 <term><option>-f <replaceable>filename</replaceable></option></term>
146 <term><option>--file=<replaceable>filename</replaceable></option></term>
147 <listitem>
148 <para>
149 Specify output file for generated script, or for the listing
150 when used with <option>-l</option>. Default is the standard
151 output.
152 </para>
153 </listitem>
154 </varlistentry>
156 <varlistentry>
157 <term><option>-F <replaceable class="parameter">format</replaceable></option></term>
158 <term><option>--format=<replaceable class="parameter">format</replaceable></option></term>
159 <listitem>
160 <para>
161 Specify format of the archive. It is not necessary to specify
162 the format, since <application>pg_restore</application> will
163 determine the format automatically. If specified, it can be
164 one of the following:
166 <variablelist>
167 <varlistentry>
168 <term><literal>t</></term>
169 <term><literal>tar</></term>
170 <listitem>
171 <para>
172 The archive is a <command>tar</command> archive. Using this
173 archive format allows reordering and/or exclusion of schema
174 elements at the time the database is restored. It is also
175 possible to limit which data is reloaded at restore time.
176 </para>
177 </listitem>
178 </varlistentry>
180 <varlistentry>
181 <term><literal>c</></term>
182 <term><literal>custom</></term>
183 <listitem>
184 <para>
185 The archive is in the custom format of
186 <application>pg_dump</application>. This is the most
187 flexible format in that it allows reordering of data load
188 as well as schema elements. This format is also compressed
189 by default.
190 </para>
191 </listitem>
192 </varlistentry>
193 </variablelist>
194 </para>
195 </listitem>
196 </varlistentry>
198 <varlistentry>
199 <term><option>-i</option></term>
200 <term><option>--ignore-version</option></term>
201 <listitem>
202 <para>
203 A deprecated option that is now ignored.
204 </para>
205 </listitem>
206 </varlistentry>
208 <varlistentry>
209 <term><option>-I <replaceable class="parameter">index</replaceable></option></term>
210 <term><option>--index=<replaceable class="parameter">index</replaceable></option></term>
211 <listitem>
212 <para>
213 Restore definition of named index only.
214 </para>
215 </listitem>
216 </varlistentry>
218 <varlistentry>
219 <term><option>-j <replaceable class="parameter">number-of-jobs</replaceable></option></term>
220 <term><option>--jobs=<replaceable class="parameter">number-of-jobs</replaceable></option></term>
221 <listitem>
222 <para>
223 Run the most time-consuming parts
224 of <application>pg_restore</> &mdash; those which load data,
225 create indexes, or create constraints &mdash; using multiple
226 concurrent jobs. This option can dramatically reduce the time
227 to restore a large database to a server running on a
228 multi-processor machine.
229 </para>
231 <para>
232 Each job is one process or one thread, depending on the
233 operating system, and uses a separate connection to the
234 server.
235 </para>
237 <para>
238 The optimal value for this option depends on the hardware
239 setup of the server, of the client, and of the network.
240 Factors include the number of CPU cores and the disk setup. A
241 good place to start is the number of CPU cores on the server,
242 but values larger than that can also lead to faster restore
243 times in many cases. Of course, values that are too high will
244 lead to decreasing performance because of thrashing.
245 </para>
247 <para>
248 Only the custom archive format is supported with this option.
249 The input file must be a regular file (not, for example, a
250 pipe). This option is ignored when emitting a script rather
251 than connecting directly to a database server. Also, multiple
252 jobs cannot be used together with the
253 option <option>--single-transaction</option>.
254 </para>
255 </listitem>
256 </varlistentry>
258 <varlistentry>
259 <term><option>-l</option></term>
260 <term><option>--list</option></term>
261 <listitem>
262 <para>
263 List the contents of the archive. The output of this operation
264 can be used with the <option>-L</option> option to restrict
265 and reorder the items that are restored.
266 </para>
267 </listitem>
268 </varlistentry>
270 <varlistentry>
271 <term><option>-L <replaceable class="parameter">list-file</replaceable></option></term>
272 <term><option>--use-list=<replaceable class="parameter">list-file</replaceable></option></term>
273 <listitem>
274 <para>
275 Restore elements in <replaceable class="PARAMETER">
276 list-file</replaceable> only, and in the
277 order they appear in the file. Lines can be moved and can also
278 be commented out by placing a <literal>;</literal> at the
279 start of the line. (See below for examples.)
280 </para>
281 </listitem>
282 </varlistentry>
284 <varlistentry>
285 <term><option>-n <replaceable class="parameter">namespace</replaceable></option></term>
286 <term><option>--schema=<replaceable class="parameter">schema</replaceable></option></term>
287 <listitem>
288 <para>
289 Restore only objects that are in the named schema. This can be
290 combined with the <option>-t</option> option to restore just a
291 specific table.
292 </para>
293 </listitem>
294 </varlistentry>
296 <varlistentry>
297 <term><option>-O</option></term>
298 <term><option>--no-owner</option></term>
299 <listitem>
300 <para>
301 Do not output commands to set
302 ownership of objects to match the original database.
303 By default, <application>pg_restore</application> issues
304 <command>ALTER OWNER</> or
305 <command>SET SESSION AUTHORIZATION</command>
306 statements to set ownership of created schema elements.
307 These statements will fail unless the initial connection to the
308 database is made by a superuser
309 (or the same user that owns all of the objects in the script).
310 With <option>-O</option>, any user name can be used for the
311 initial connection, and this user will own all the created objects.
312 </para>
313 </listitem>
314 </varlistentry>
316 <varlistentry>
317 <term><option>--no-tablespaces</option></term>
318 <listitem>
319 <para>
320 Do not output commands to select tablespaces.
321 With this option, all objects will be created in whichever
322 tablespace is the default during restore.
323 </para>
324 </listitem>
325 </varlistentry>
327 <varlistentry>
328 <term><option>-P <replaceable class="parameter">function-name(argtype [, ...])</replaceable></option></term>
329 <term><option>--function=<replaceable class="parameter">function-name(argtype [, ...])</replaceable></option></term>
330 <listitem>
331 <para>
332 Restore the named function only. Be careful to spell the function
333 name and arguments exactly as they appear in the dump file's table
334 of contents.
335 </para>
336 </listitem>
337 </varlistentry>
339 <varlistentry>
340 <term><option>-R</option></term>
341 <term><option>--no-reconnect</option></term>
342 <listitem>
343 <para>
344 This option is obsolete but still accepted for backwards
345 compatibility.
346 </para>
347 </listitem>
348 </varlistentry>
350 <varlistentry>
351 <term><option>-s</option></term>
352 <term><option>--schema-only</option></term>
353 <listitem>
354 <para>
355 Restore only the schema (data definitions), not the data (table
356 contents). Sequence current values will not be restored, either.
357 (Do not confuse this with the <option>--schema</> option, which
358 uses the word <quote>schema</> in a different meaning.)
359 </para>
360 </listitem>
361 </varlistentry>
363 <varlistentry>
364 <term><option>-S <replaceable class="parameter">username</replaceable></option></term>
365 <term><option>--superuser=<replaceable class="parameter">username</replaceable></option></term>
366 <listitem>
367 <para>
368 Specify the superuser user name to use when disabling triggers.
369 This is only relevant if <option>--disable-triggers</> is used.
370 </para>
371 </listitem>
372 </varlistentry>
374 <varlistentry>
375 <term><option>-t <replaceable class="parameter">table</replaceable></option></term>
376 <term><option>--table=<replaceable class="parameter">table</replaceable></option></term>
377 <listitem>
378 <para>
379 Restore definition and/or data of named table only.
380 </para>
381 </listitem>
382 </varlistentry>
384 <varlistentry>
385 <term><option>-T <replaceable class="parameter">trigger</replaceable></option></term>
386 <term><option>--trigger=<replaceable class="parameter">trigger</replaceable></option></term>
387 <listitem>
388 <para>
389 Restore named trigger only.
390 </para>
391 </listitem>
392 </varlistentry>
394 <varlistentry>
395 <term><option>-v</option></term>
396 <term><option>--verbose</option></term>
397 <listitem>
398 <para>
399 Specifies verbose mode.
400 </para>
401 </listitem>
402 </varlistentry>
404 <varlistentry>
405 <term><option>-x</option></term>
406 <term><option>--no-privileges</option></term>
407 <term><option>--no-acl</option></term>
408 <listitem>
409 <para>
410 Prevent restoration of access privileges (grant/revoke commands).
411 </para>
412 </listitem>
413 </varlistentry>
415 <varlistentry>
416 <term><option>--disable-triggers</></term>
417 <listitem>
418 <para>
419 This option is only relevant when performing a data-only restore.
420 It instructs <application>pg_restore</application> to execute commands
421 to temporarily disable triggers on the target tables while
422 the data is reloaded. Use this if you have referential
423 integrity checks or other triggers on the tables that you
424 do not want to invoke during data reload.
425 </para>
427 <para>
428 Presently, the commands emitted for
429 <option>--disable-triggers</> must be done as superuser. So, you
430 should also specify a superuser name with <option>-S</>, or
431 preferably run <application>pg_restore</application> as a
432 <productname>PostgreSQL</> superuser.
433 </para>
434 </listitem>
435 </varlistentry>
437 <varlistentry>
438 <term><option>--use-set-session-authorization</option></term>
439 <listitem>
440 <para>
441 Output SQL-standard <command>SET SESSION AUTHORIZATION</> commands
442 instead of <command>ALTER OWNER</> commands to determine object
443 ownership. This makes the dump more standards compatible, but
444 depending on the history of the objects in the dump, might not restore
445 properly.
446 </para>
447 </listitem>
448 </varlistentry>
450 <varlistentry>
451 <term><option>--no-data-for-failed-tables</option></term>
452 <listitem>
453 <para>
454 By default, table data is restored even if the creation command
455 for the table failed (e.g., because it already exists).
456 With this option, data for such a table is skipped.
457 This behavior is useful if the target database already
458 contains the desired table contents. For example,
459 auxiliary tables for <productname>PostgreSQL</> extensions
460 such as <productname>PostGIS</> might already be loaded in
461 the target database; specifying this option prevents duplicate
462 or obsolete data from being loaded into them.
463 </para>
465 <para>
466 This option is effective only when restoring directly into a
467 database, not when producing SQL script output.
468 </para>
469 </listitem>
470 </varlistentry>
472 <varlistentry>
473 <term><option>-1</option></term>
474 <term><option>--single-transaction</option></term>
475 <listitem>
476 <para>
477 Execute the restore as a single transaction (that is, wrap the
478 emitted commands in <command>BEGIN</>/<command>COMMIT</>). This
479 ensures that either all the commands complete successfully, or no
480 changes are applied. This option implies
481 <option>--exit-on-error</>.
482 </para>
483 </listitem>
484 </varlistentry>
486 </variablelist>
487 </para>
489 <para>
490 <application>pg_restore</application> also accepts
491 the following command line arguments for connection parameters:
493 <variablelist>
494 <varlistentry>
495 <term><option>-h <replaceable class="parameter">host</replaceable></option></term>
496 <term><option>--host=<replaceable class="parameter">host</replaceable></option></term>
497 <listitem>
498 <para>
499 Specifies the host name of the machine on which the server is
500 running. If the value begins with a slash, it is used as the
501 directory for the Unix domain socket. The default is taken
502 from the <envar>PGHOST</envar> environment variable, if set,
503 else a Unix domain socket connection is attempted.
504 </para>
505 </listitem>
506 </varlistentry>
508 <varlistentry>
509 <term><option>-p <replaceable class="parameter">port</replaceable></option></term>
510 <term><option>--port=<replaceable class="parameter">port</replaceable></option></term>
511 <listitem>
512 <para>
513 Specifies the TCP port or local Unix domain socket file
514 extension on which the server is listening for connections.
515 Defaults to the <envar>PGPORT</envar> environment variable, if
516 set, or a compiled-in default.
517 </para>
518 </listitem>
519 </varlistentry>
521 <varlistentry>
522 <term><option>-U <replaceable>username</replaceable></option></term>
523 <term><option>--username=<replaceable class="parameter">username</replaceable></option></term>
524 <listitem>
525 <para>
526 User name to connect as.
527 </para>
528 </listitem>
529 </varlistentry>
531 <varlistentry>
532 <term><option>-w</></term>
533 <term><option>--no-password</></term>
534 <listitem>
535 <para>
536 Never issue a password prompt. If the server requires
537 password authentication and a password is not available by
538 other means such as a <filename>.pgpass</filename> file, the
539 connection attempt will fail. This option can be useful in
540 batch jobs and scripts where no user is present to enter a
541 password.
542 </para>
543 </listitem>
544 </varlistentry>
546 <varlistentry>
547 <term><option>-W</option></term>
548 <term><option>--password</option></term>
549 <listitem>
550 <para>
551 Force <application>pg_restore</application> to prompt for a
552 password before connecting to a database.
553 </para>
555 <para>
556 This option is never essential, since
557 <application>pg_restore</application> will automatically prompt
558 for a password if the server demands password authentication.
559 However, <application>pg_restore</application> will waste a
560 connection attempt finding out that the server wants a password.
561 In some cases it is worth typing <option>-W</> to avoid the extra
562 connection attempt.
563 </para>
564 </listitem>
565 </varlistentry>
567 <varlistentry>
568 <term><option>--role=<replaceable class="parameter">rolename</replaceable></option></term>
569 <listitem>
570 <para>
571 Specifies a role name to be used to perform the restore.
572 This option causes <application>pg_restore</> to issue a
573 <command>SET ROLE</> <replaceable class="parameter">rolename</>
574 command after connecting to the database. It is useful when the
575 authenticated user (specified by <option>-U</>) lacks privileges
576 needed by <application>pg_restore</>, but can switch to a role with
577 the required rights. Some installations have a policy against
578 logging in directly as a superuser, and use of this option allows
579 restores to be performed without violating the policy.
580 </para>
581 </listitem>
582 </varlistentry>
584 </variablelist>
585 </para>
586 </refsect1>
589 <refsect1>
590 <title>Environment</title>
592 <variablelist>
593 <varlistentry>
594 <term><envar>PGHOST</envar></term>
595 <term><envar>PGOPTIONS</envar></term>
596 <term><envar>PGPORT</envar></term>
597 <term><envar>PGUSER</envar></term>
599 <listitem>
600 <para>
601 Default connection parameters
602 </para>
603 </listitem>
604 </varlistentry>
605 </variablelist>
607 <para>
608 This utility, like most other <productname>PostgreSQL</> utilities,
609 also uses the environment variables supported by <application>libpq</>
610 (see <xref linkend="libpq-envars">).
611 </para>
613 </refsect1>
616 <refsect1 id="app-pgrestore-diagnostics">
617 <title>Diagnostics</title>
619 <para>
620 When a direct database connection is specified using the
621 <option>-d</option> option, <application>pg_restore</application>
622 internally executes <acronym>SQL</acronym> statements. If you have
623 problems running <application>pg_restore</application>, make sure
624 you are able to select information from the database using, for
625 example, <xref linkend="app-psql">. Also, any default connection
626 settings and environment variables used by the
627 <application>libpq</application> front-end library will apply.
628 </para>
629 </refsect1>
632 <refsect1 id="app-pgrestore-notes">
633 <title>Notes</title>
635 <para>
636 If your installation has any local additions to the
637 <literal>template1</> database, be careful to load the output of
638 <application>pg_restore</application> into a truly empty database;
639 otherwise you are likely to get errors due to duplicate definitions
640 of the added objects. To make an empty database without any local
641 additions, copy from <literal>template0</> not <literal>template1</>, for example:
642 <programlisting>
643 CREATE DATABASE foo WITH TEMPLATE template0;
644 </programlisting>
645 </para>
647 <para>
648 The limitations of <application>pg_restore</application> are detailed below.
650 <itemizedlist>
651 <listitem>
652 <para>
653 When restoring data to a pre-existing table and the option
654 <option>--disable-triggers</> is used,
655 <application>pg_restore</application> emits commands
656 to disable triggers on user tables before inserting the data then emits commands to
657 re-enable them after the data has been inserted. If the restore is stopped in the
658 middle, the system catalogs might be left in the wrong state.
659 </para>
660 </listitem>
662 <listitem>
663 <para>
664 <application>pg_restore</application> cannot restore large objects
665 selectively, for instance only those for a specific table. If
666 an archive contains large objects, then all large objects will be
667 restored, or none of them if they are excluded via <option>-L</option>,
668 <option>-t</option>, or other options.
669 </para>
670 </listitem>
672 </itemizedlist>
673 </para>
675 <para>
676 See also the <xref linkend="app-pgdump"> documentation for details on
677 limitations of <application>pg_dump</application>.
678 </para>
680 <para>
681 Once restored, it is wise to run <command>ANALYZE</> on each
682 restored table so the optimizer has useful statistics.
683 </para>
685 </refsect1>
688 <refsect1 id="app-pgrestore-examples">
689 <title>Examples</title>
691 <para>
692 Assume we have dumped a database called <literal>mydb</> into a
693 custom-format dump file:
695 <screen>
696 <prompt>$</prompt> <userinput>pg_dump -Fc mydb &gt; db.dump</userinput>
697 </screen>
698 </para>
700 <para>
701 To drop the database and recreate it from the dump:
703 <screen>
704 <prompt>$</prompt> <userinput>dropdb mydb</userinput>
705 <prompt>$</prompt> <userinput>pg_restore -C -d postgres db.dump</userinput>
706 </screen>
708 The database named in the <option>-d</> switch can be any database existing
709 in the cluster; <application>pg_restore</> only uses it to issue the
710 <command>CREATE DATABASE</> command for <literal>mydb</>. With
711 <option>-C</>, data is always restored into the database name that appears
712 in the dump file.
713 </para>
715 <para>
716 To reload the dump into a new database called <literal>newdb</>:
718 <screen>
719 <prompt>$</prompt> <userinput>createdb -T template0 newdb</userinput>
720 <prompt>$</prompt> <userinput>pg_restore -d newdb db.dump</userinput>
721 </screen>
723 Notice we don't use <option>-C</>, and instead connect directly to the
724 database to be restored into. Also note that we clone the new database
725 from <literal>template0</> not <literal>template1</>, to ensure it is
726 initially empty.
727 </para>
729 <para>
730 To reorder database items, it is first necessary to dump the table of
731 contents of the archive:
732 <screen>
733 <prompt>$</prompt> <userinput>pg_restore -l db.dump &gt; db.list</userinput>
734 </screen>
735 The listing file consists of a header and one line for each item, e.g.:
736 <programlisting>
738 ; Archive created at Fri Jul 28 22:28:36 2000
739 ; dbname: mydb
740 ; TOC Entries: 74
741 ; Compression: 0
742 ; Dump Version: 1.4-0
743 ; Format: CUSTOM
746 ; Selected TOC Entries:
748 2; 145344 TABLE species postgres
749 3; 145344 ACL species
750 4; 145359 TABLE nt_header postgres
751 5; 145359 ACL nt_header
752 6; 145402 TABLE species_records postgres
753 7; 145402 ACL species_records
754 8; 145416 TABLE ss_old postgres
755 9; 145416 ACL ss_old
756 10; 145433 TABLE map_resolutions postgres
757 11; 145433 ACL map_resolutions
758 12; 145443 TABLE hs_old postgres
759 13; 145443 ACL hs_old
760 </programlisting>
761 Semicolons start a comment, and the numbers at the start of lines refer to the
762 internal archive ID assigned to each item.
763 </para>
765 <para>
766 Lines in the file can be commented out, deleted, and reordered. For example:
767 <programlisting>
768 10; 145433 TABLE map_resolutions postgres
769 ;2; 145344 TABLE species postgres
770 ;4; 145359 TABLE nt_header postgres
771 6; 145402 TABLE species_records postgres
772 ;8; 145416 TABLE ss_old postgres
773 </programlisting>
774 could be used as input to <application>pg_restore</application> and would only restore
775 items 10 and 6, in that order:
776 <screen>
777 <prompt>$</prompt> <userinput>pg_restore -L db.list db.dump</userinput>
778 </screen>
779 </para>
781 </refsect1>
783 <refsect1>
784 <title>See Also</title>
786 <simplelist type="inline">
787 <member><xref linkend="app-pgdump"></member>
788 <member><xref linkend="app-pg-dumpall"></member>
789 <member><xref linkend="app-psql"></member>
790 </simplelist>
791 </refsect1>
792 </refentry>