Debian Codegraph @

doc/src/sgml/client-auth.sgml

raw
1 <!-- doc/src/sgml/client-auth.sgml -->
2
3 <chapter id="client-authentication">
4 <title>Client Authentication</title>
5
6 <indexterm zone="client-authentication">
7 <primary>client authentication</primary>
8 </indexterm>
9
10 <para>
11 When a client application connects to the database server, it
12 specifies which <productname>PostgreSQL</productname> database user name it
13 wants to connect as, much the same way one logs into a Unix computer
14 as a particular user. Within the SQL environment the active database
15 user name determines access privileges to database objects &mdash; see
16 <xref linkend="user-manag"/> for more information. Therefore, it is
17 essential to restrict which database users can connect.
18 </para>
19
20 <note>
21 <para>
22 As explained in <xref linkend="user-manag"/>,
23 <productname>PostgreSQL</productname> actually does privilege
24 management in terms of <quote>roles</quote>. In this chapter, we
25 consistently use <firstterm>database user</firstterm> to mean <quote>role with the
26 <literal>LOGIN</literal> privilege</quote>.
27 </para>
28 </note>
29
30 <para>
31 <firstterm>Authentication</firstterm> is the process by which the
32 database server establishes the identity of the client, and by
33 extension determines whether the client application (or the user
34 who runs the client application) is permitted to connect with the
35 database user name that was requested.
36 </para>
37
38 <para>
39 <productname>PostgreSQL</productname> offers a number of different
40 client authentication methods. The method used to authenticate a
41 particular client connection can be selected on the basis of
42 (client) host address, database, and user.
43 </para>
44
45 <para>
46 <productname>PostgreSQL</productname> database user names are logically
47 separate from user names of the operating system in which the server
48 runs. If all the users of a particular server also have accounts on
49 the server's machine, it makes sense to assign database user names
50 that match their operating system user names. However, a server that
51 accepts remote connections might have many database users who have no local
52 operating system
53 account, and in such cases there need be no connection between
54 database user names and OS user names.
55 </para>
56
57 <sect1 id="auth-pg-hba-conf">
58 <title>The <filename>pg_hba.conf</filename> File</title>
59
60 <indexterm zone="auth-pg-hba-conf">
61 <primary>pg_hba.conf</primary>
62 </indexterm>
63
64 <para>
65 Client authentication is controlled by a configuration file,
66 which traditionally is named
67 <filename>pg_hba.conf</filename> and is stored in the database
68 cluster's data directory.
69 (<acronym>HBA</acronym> stands for host-based authentication.) A default
70 <filename>pg_hba.conf</filename> file is installed when the data
71 directory is initialized by <xref linkend="app-initdb"/>. It is
72 possible to place the authentication configuration file elsewhere,
73 however; see the <xref linkend="guc-hba-file"/> configuration parameter.
74 </para>
75
76 <para>
77 The <filename>pg_hba.conf</filename> file is read on start-up and when
78 the main server process receives a
79 <systemitem>SIGHUP</systemitem><indexterm><primary>SIGHUP</primary></indexterm>
80 signal. If you edit the file on an
81 active system, you will need to signal the postmaster
82 (using <literal>pg_ctl reload</literal>, calling the SQL function
83 <function>pg_reload_conf()</function>, or using <literal>kill
84 -HUP</literal>) to make it re-read the file.
85 </para>
86
87 <note>
88 <para>
89 The preceding statement is not true on Microsoft Windows: there, any
90 changes in the <filename>pg_hba.conf</filename> file are immediately
91 applied by subsequent new connections.
92 </para>
93 </note>
94
95 <para>
96 The system view
97 <link linkend="view-pg-hba-file-rules"><structname>pg_hba_file_rules</structname></link>
98 can be helpful for pre-testing changes to the <filename>pg_hba.conf</filename>
99 file, or for diagnosing problems if loading of the file did not have the
100 desired effects. Rows in the view with
101 non-null <structfield>error</structfield> fields indicate problems in the
102 corresponding lines of the file.
103 </para>
104
105 <para>
106 The general format of the <filename>pg_hba.conf</filename> file is
107 a set of records, one per line. Blank lines are ignored, as is any
108 text after the <literal>#</literal> comment character.
109 A record can be continued onto the next line by ending the line with
110 a backslash. (Backslashes are not special except at the end of a line.)
111 A record is made
112 up of a number of fields which are separated by spaces and/or tabs.
113 Fields can contain white space if the field value is double-quoted.
114 Quoting one of the keywords in a database, user, or address field (e.g.,
115 <literal>all</literal> or <literal>replication</literal>) makes the word lose its special
116 meaning, and just match a database, user, or host with that name.
117 Backslash line continuation applies even within quoted text or comments.
118 </para>
119
120 <para>
121 Each authentication record specifies a connection type, a client IP address
122 range (if relevant for the connection type), a database name, a user name,
123 and the authentication method to be used for connections matching
124 these parameters. The first record with a matching connection type,
125 client address, requested database, and user name is used to perform
126 authentication. There is no <quote>fall-through</quote> or
127 <quote>backup</quote>: if one record is chosen and the authentication
128 fails, subsequent records are not considered. If no record matches,
129 access is denied.
130 </para>
131
132 <para>
133 Each record can be an include directive or an authentication record.
134 Include directives specify files that can be included, that contain
135 additional records. The records will be inserted in place of the
136 include directives. Include directives only contain two fields:
137 <literal>include</literal>, <literal>include_if_exists</literal> or
138 <literal>include_dir</literal> directive and the file or directory to be
139 included. The file or directory can be a relative or absolute path, and can
140 be double-quoted. For the <literal>include_dir</literal> form, all files
141 not starting with a <literal>.</literal> and ending with
142 <literal>.conf</literal> will be included. Multiple files within an include
143 directory are processed in file name order (according to C locale rules,
144 i.e., numbers before letters, and uppercase letters before lowercase ones).
145 </para>
146
147 <para>
148 A record can have several formats:
149 <synopsis>
150 local <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
151 host <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>address</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
152 hostssl <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>address</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
153 hostnossl <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>address</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
154 hostgssenc <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>address</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
155 hostnogssenc <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>address</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
156 host <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>IP-address</replaceable> <replaceable>IP-mask</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
157 hostssl <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>IP-address</replaceable> <replaceable>IP-mask</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
158 hostnossl <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>IP-address</replaceable> <replaceable>IP-mask</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
159 hostgssenc <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>IP-address</replaceable> <replaceable>IP-mask</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
160 hostnogssenc <replaceable>database</replaceable> <replaceable>user</replaceable> <replaceable>IP-address</replaceable> <replaceable>IP-mask</replaceable> <replaceable>auth-method</replaceable> <optional><replaceable>auth-options</replaceable></optional>
161 include <replaceable>file</replaceable>
162 include_if_exists <replaceable>file</replaceable>
163 include_dir <replaceable>directory</replaceable>
164 </synopsis>
165 The meaning of the fields is as follows:
166
167 <variablelist>
168 <varlistentry>
169 <term><literal>local</literal></term>
170 <listitem>
171 <para>
172 This record matches connection attempts using Unix-domain
173 sockets. Without a record of this type, Unix-domain socket
174 connections are disallowed.
175 </para>
176 </listitem>
177 </varlistentry>
178
179 <varlistentry>
180 <term><literal>host</literal></term>
181 <listitem>
182 <para>
183 This record matches connection attempts made using TCP/IP.
184 <literal>host</literal> records match
185 <acronym>SSL</acronym> or non-<acronym>SSL</acronym> connection
186 attempts as well as <acronym>GSSAPI</acronym> encrypted or
187 non-<acronym>GSSAPI</acronym> encrypted connection attempts.
188 </para>
189 <note>
190 <para>
191 Remote TCP/IP connections will not be possible unless
192 the server is started with an appropriate value for the
193 <xref linkend="guc-listen-addresses"/> configuration parameter,
194 since the default behavior is to listen for TCP/IP connections
195 only on the local loopback address <literal>localhost</literal>.
196 </para>
197 </note>
198 </listitem>
199 </varlistentry>
200
201 <varlistentry>
202 <term><literal>hostssl</literal></term>
203 <listitem>
204 <para>
205 This record matches connection attempts made using TCP/IP,
206 but only when the connection is made with <acronym>SSL</acronym>
207 encryption.
208 </para>
209
210 <para>
211 To make use of this option the server must be built with
212 <acronym>SSL</acronym> support. Furthermore,
213 <acronym>SSL</acronym> must be enabled
214 by setting the <xref linkend="guc-ssl"/> configuration parameter (see
215 <xref linkend="ssl-tcp"/> for more information).
216 Otherwise, the <literal>hostssl</literal> record is ignored except for
217 logging a warning that it cannot match any connections.
218 </para>
219 </listitem>
220 </varlistentry>
221
222 <varlistentry>
223 <term><literal>hostnossl</literal></term>
224 <listitem>
225 <para>
226 This record type has the opposite behavior of <literal>hostssl</literal>;
227 it only matches connection attempts made over
228 TCP/IP that do not use <acronym>SSL</acronym>.
229 </para>
230 </listitem>
231 </varlistentry>
232
233 <varlistentry>
234 <term><literal>hostgssenc</literal></term>
235 <listitem>
236 <para>
237 This record matches connection attempts made using TCP/IP,
238 but only when the connection is made with <acronym>GSSAPI</acronym>
239 encryption.
240 </para>
241
242 <para>
243 To make use of this option the server must be built with
244 <acronym>GSSAPI</acronym> support. Otherwise,
245 the <literal>hostgssenc</literal> record is ignored except for logging
246 a warning that it cannot match any connections.
247 </para>
248 </listitem>
249 </varlistentry>
250
251 <varlistentry>
252 <term><literal>hostnogssenc</literal></term>
253 <listitem>
254 <para>
255 This record type has the opposite behavior of <literal>hostgssenc</literal>;
256 it only matches connection attempts made over
257 TCP/IP that do not use <acronym>GSSAPI</acronym> encryption.
258 </para>
259 </listitem>
260 </varlistentry>
261
262 <varlistentry>
263 <term><replaceable>database</replaceable></term>
264 <listitem>
265 <para>
266 Specifies which database name(s) this record matches. The value
267 <literal>all</literal> specifies that it matches all databases.
268 The value <literal>sameuser</literal> specifies that the record
269 matches if the requested database has the same name as the
270 requested user. The value <literal>samerole</literal> specifies that
271 the requested user must be a member of the role with the same
272 name as the requested database. (<literal>samegroup</literal> is an
273 obsolete but still accepted spelling of <literal>samerole</literal>.)
274 Superusers are not considered to be members of a role for the
275 purposes of <literal>samerole</literal> unless they are explicitly
276 members of the role, directly or indirectly, and not just by
277 virtue of being a superuser.
278 The value <literal>replication</literal> specifies that the record
279 matches if a physical replication connection is requested, however, it
280 doesn't match with logical replication connections. Note that physical
281 replication connections do not specify any particular database whereas
282 logical replication connections do specify it.
283 Otherwise, this is the name of a specific
284 <productname>PostgreSQL</productname> database or a regular expression.
285 Multiple database names and/or regular expressions can be supplied by
286 separating them with commas.
287 </para>
288 <para>
289 If the database name starts with a slash (<literal>/</literal>), the
290 remainder of the name is treated as a regular expression.
291 (See <xref linkend="posix-syntax-details"/> for details of
292 <productname>PostgreSQL</productname>'s regular expression syntax.)
293 </para>
294 <para>
295 A separate file containing database names and/or regular expressions
296 can be specified by preceding the file name with <literal>@</literal>.
297 </para>
298 </listitem>
299 </varlistentry>
300
301 <varlistentry>
302 <term><replaceable>user</replaceable></term>
303 <listitem>
304 <para>
305 Specifies which database user name(s) this record
306 matches. The value <literal>all</literal> specifies that it
307 matches all users. Otherwise, this is either the name of a specific
308 database user, a regular expression (when starting with a slash
309 (<literal>/</literal>), or a group name preceded by <literal>+</literal>.
310 (Recall that there is no real distinction between users and groups
311 in <productname>PostgreSQL</productname>; a <literal>+</literal> mark really means
312 <quote>match any of the roles that are directly or indirectly members
313 of this role</quote>, while a name without a <literal>+</literal> mark matches
314 only that specific role.) For this purpose, a superuser is only
315 considered to be a member of a role if they are explicitly a member
316 of the role, directly or indirectly, and not just by virtue of
317 being a superuser.
318 Multiple user names and/or regular expressions can be supplied by
319 separating them with commas.
320 </para>
321 <para>
322 If the user name starts with a slash (<literal>/</literal>), the
323 remainder of the name is treated as a regular expression.
324 (See <xref linkend="posix-syntax-details"/> for details of
325 <productname>PostgreSQL</productname>'s regular expression syntax.)
326 </para>
327 <para>
328 A separate file containing user names and/or regular expressions can
329 be specified by preceding the file name with <literal>@</literal>.
330 </para>
331 </listitem>
332 </varlistentry>
333
334 <varlistentry>
335 <term><replaceable>address</replaceable></term>
336 <listitem>
337 <para>
338 Specifies the client machine address(es) that this record
339 matches. This field can contain either a host name, an IP
340 address range, or one of the special key words mentioned below.
341 </para>
342
343 <para>
344 An IP address range is specified using standard numeric notation
345 for the range's starting address, then a slash (<literal>/</literal>)
346 and a <acronym>CIDR</acronym> mask length. The mask
347 length indicates the number of high-order bits of the client
348 IP address that must match. Bits to the right of this should
349 be zero in the given IP address.
350 There must not be any white space between the IP address, the
351 <literal>/</literal>, and the CIDR mask length.
352 </para>
353
354 <para>
355 Typical examples of an IPv4 address range specified this way are
356 <literal>172.20.143.89/32</literal> for a single host, or
357 <literal>172.20.143.0/24</literal> for a small network, or
358 <literal>10.6.0.0/16</literal> for a larger one.
359 An IPv6 address range might look like <literal>::1/128</literal>
360 for a single host (in this case the IPv6 loopback address) or
361 <literal>fe80::7a31:c1ff:0000:0000/96</literal> for a small
362 network.
363 <literal>0.0.0.0/0</literal> represents all
364 IPv4 addresses, and <literal>::0/0</literal> represents
365 all IPv6 addresses.
366 To specify a single host, use a mask length of 32 for IPv4 or
367 128 for IPv6. In a network address, do not omit trailing zeroes.
368 </para>
369
370 <para>
371 An entry given in IPv4 format will match only IPv4 connections,
372 and an entry given in IPv6 format will match only IPv6 connections,
373 even if the represented address is in the IPv4-in-IPv6 range.
374 </para>
375
376 <para>
377 You can also write <literal>all</literal> to match any IP address,
378 <literal>samehost</literal> to match any of the server's own IP
379 addresses, or <literal>samenet</literal> to match any address in any
380 subnet that the server is directly connected to.
381 </para>
382
383 <para>
384 If a host name is specified (anything that is not an IP address
385 range or a special key word is treated as a host name),
386 that name is compared with the result of a reverse name
387 resolution of the client's IP address (e.g., reverse DNS
388 lookup, if DNS is used). Host name comparisons are case
389 insensitive. If there is a match, then a forward name
390 resolution (e.g., forward DNS lookup) is performed on the host
391 name to check whether any of the addresses it resolves to are
392 equal to the client's IP address. If both directions match,
393 then the entry is considered to match. (The host name that is
394 used in <filename>pg_hba.conf</filename> should be the one that
395 address-to-name resolution of the client's IP address returns,
396 otherwise the line won't be matched. Some host name databases
397 allow associating an IP address with multiple host names, but
398 the operating system will only return one host name when asked
399 to resolve an IP address.)
400 </para>
401
402 <para>
403 A host name specification that starts with a dot
404 (<literal>.</literal>) matches a suffix of the actual host
405 name. So <literal>.example.com</literal> would match
406 <literal>foo.example.com</literal> (but not just
407 <literal>example.com</literal>).
408 </para>
409
410 <para>
411 When host names are specified
412 in <filename>pg_hba.conf</filename>, you should make sure that
413 name resolution is reasonably fast. It can be of advantage to
414 set up a local name resolution cache such
415 as <command>nscd</command>. Also, you may wish to enable the
416 configuration parameter <varname>log_hostname</varname> to see
417 the client's host name instead of the IP address in the log.
418 </para>
419
420 <para>
421 These fields do not apply to <literal>local</literal> records.
422 </para>
423
424 <note>
425 <para>
426 Users sometimes wonder why host names are handled
427 in this seemingly complicated way, with two name resolutions
428 including a reverse lookup of the client's IP address. This
429 complicates use of the feature in case the client's reverse DNS
430 entry is not set up or yields some undesirable host name.
431 It is done primarily for efficiency: this way, a connection attempt
432 requires at most two resolver lookups, one reverse and one forward.
433 If there is a resolver problem with some address, it becomes only
434 that client's problem. A hypothetical alternative
435 implementation that only did forward lookups would have to
436 resolve every host name mentioned in
437 <filename>pg_hba.conf</filename> during every connection attempt.
438 That could be quite slow if many names are listed.
439 And if there is a resolver problem with one of the host names,
440 it becomes everyone's problem.
441 </para>
442
443 <para>
444 Also, a reverse lookup is necessary to implement the suffix
445 matching feature, because the actual client host name needs to
446 be known in order to match it against the pattern.
447 </para>
448
449 <para>
450 Note that this behavior is consistent with other popular
451 implementations of host name-based access control, such as the
452 Apache HTTP Server and TCP Wrappers.
453 </para>
454 </note>
455 </listitem>
456 </varlistentry>
457
458 <varlistentry>
459 <term><replaceable>IP-address</replaceable></term>
460 <term><replaceable>IP-mask</replaceable></term>
461 <listitem>
462 <para>
463 These two fields can be used as an alternative to the
464 <replaceable>IP-address</replaceable><literal>/</literal><replaceable>mask-length</replaceable>
465 notation. Instead of
466 specifying the mask length, the actual mask is specified in a
467 separate column. For example, <literal>255.0.0.0</literal> represents an IPv4
468 CIDR mask length of 8, and <literal>255.255.255.255</literal> represents a
469 CIDR mask length of 32.
470 </para>
471
472 <para>
473 These fields do not apply to <literal>local</literal> records.
474 </para>
475 </listitem>
476 </varlistentry>
477
478 <varlistentry>
479 <term><replaceable>auth-method</replaceable></term>
480 <listitem>
481 <para>
482 Specifies the authentication method to use when a connection matches
483 this record. The possible choices are summarized here; details
484 are in <xref linkend="auth-methods"/>. All the options
485 are lower case and treated case sensitively, so even acronyms like
486 <literal>ldap</literal> must be specified as lower case.
487
488 <variablelist>
489 <varlistentry>
490 <term><literal>trust</literal></term>
491 <listitem>
492 <para>
493 Allow the connection unconditionally. This method
494 allows anyone that can connect to the
495 <productname>PostgreSQL</productname> database server to login as
496 any <productname>PostgreSQL</productname> user they wish,
497 without the need for a password or any other authentication. See <xref
498 linkend="auth-trust"/> for details.
499 </para>
500 </listitem>
501 </varlistentry>
502
503 <varlistentry>
504 <term><literal>reject</literal></term>
505 <listitem>
506 <para>
507 Reject the connection unconditionally. This is useful for
508 <quote>filtering out</quote> certain hosts from a group, for example a
509 <literal>reject</literal> line could block a specific host from connecting,
510 while a later line allows the remaining hosts in a specific
511 network to connect.
512 </para>
513 </listitem>
514 </varlistentry>
515
516 <varlistentry>
517 <term><literal>scram-sha-256</literal></term>
518 <listitem>
519 <para>
520 Perform SCRAM-SHA-256 authentication to verify the user's
521 password. See <xref linkend="auth-password"/> for details.
522 </para>
523 </listitem>
524 </varlistentry>
525
526 <varlistentry>
527 <term><literal>md5</literal></term>
528 <listitem>
529 <para>
530 Perform SCRAM-SHA-256 or MD5 authentication to verify the
531 user's password. See <xref linkend="auth-password"/>
532 for details.
533 </para>
534 <warning>
535 <para>
536 Support for MD5-encrypted passwords is deprecated and will be
537 removed in a future release of
538 <productname>PostgreSQL</productname>. Refer to
539 <xref linkend="auth-password"/> for details about migrating to
540 another password type.
541 </para>
542 </warning>
543 </listitem>
544 </varlistentry>
545
546 <varlistentry>
547 <term><literal>password</literal></term>
548 <listitem>
549 <para>
550 Require the client to supply an unencrypted password for
551 authentication.
552 Since the password is sent in clear text over the
553 network, this should not be used on untrusted networks.
554 See <xref linkend="auth-password"/> for details.
555 </para>
556 </listitem>
557 </varlistentry>
558
559 <varlistentry>
560 <term><literal>gss</literal></term>
561 <listitem>
562 <para>
563 Use GSSAPI to authenticate the user. This is only
564 available for TCP/IP connections. See <xref
565 linkend="gssapi-auth"/> for details. It can be used in conjunction
566 with GSSAPI encryption.
567 </para>
568 </listitem>
569 </varlistentry>
570
571 <varlistentry>
572 <term><literal>sspi</literal></term>
573 <listitem>
574 <para>
575 Use SSPI to authenticate the user. This is only
576 available on Windows. See <xref
577 linkend="sspi-auth"/> for details.
578 </para>
579 </listitem>
580 </varlistentry>
581
582 <varlistentry>
583 <term><literal>ident</literal></term>
584 <listitem>
585 <para>
586 Obtain the operating system user name of the client
587 by contacting the ident server on the client
588 and check if it matches the requested database user name.
589 Ident authentication can only be used on TCP/IP
590 connections. When specified for local connections, peer
591 authentication will be used instead.
592 See <xref linkend="auth-ident"/> for details.
593 </para>
594 </listitem>
595 </varlistentry>
596
597 <varlistentry>
598 <term><literal>peer</literal></term>
599 <listitem>
600 <para>
601 Obtain the client's operating system user name from the operating
602 system and check if it matches the requested database user name.
603 This is only available for local connections.
604 See <xref linkend="auth-peer"/> for details.
605 </para>
606 </listitem>
607 </varlistentry>
608
609 <varlistentry>
610 <term><literal>ldap</literal></term>
611 <listitem>
612 <para>
613 Authenticate using an <acronym>LDAP</acronym> server. See <xref
614 linkend="auth-ldap"/> for details.
615 </para>
616 </listitem>
617 </varlistentry>
618
619 <varlistentry>
620 <term><literal>radius</literal></term>
621 <listitem>
622 <para>
623 Authenticate using a RADIUS server. See <xref
624 linkend="auth-radius"/> for details.
625 </para>
626 </listitem>
627 </varlistentry>
628
629 <varlistentry>
630 <term><literal>cert</literal></term>
631 <listitem>
632 <para>
633 Authenticate using SSL client certificates. See
634 <xref linkend="auth-cert"/> for details.
635 </para>
636 </listitem>
637 </varlistentry>
638
639 <varlistentry>
640 <term><literal>pam</literal></term>
641 <listitem>
642 <para>
643 Authenticate using the Pluggable Authentication Modules
644 (PAM) service provided by the operating system. See <xref
645 linkend="auth-pam"/> for details.
646 </para>
647 </listitem>
648 </varlistentry>
649
650 <varlistentry>
651 <term><literal>bsd</literal></term>
652 <listitem>
653 <para>
654 Authenticate using the BSD Authentication service provided by the
655 operating system. See <xref linkend="auth-bsd"/> for details.
656 </para>
657 </listitem>
658 </varlistentry>
659
660 <varlistentry>
661 <term><literal>oauth</literal></term>
662 <listitem>
663 <para>
664 Authorize and optionally authenticate using a third-party OAuth 2.0
665 identity provider. See <xref linkend="auth-oauth"/> for details.
666 </para>
667 </listitem>
668 </varlistentry>
669 </variablelist>
670
671 </para>
672 </listitem>
673 </varlistentry>
674
675 <varlistentry>
676 <term><replaceable>auth-options</replaceable></term>
677 <listitem>
678 <para>
679 After the <replaceable>auth-method</replaceable> field, there can be field(s) of
680 the form <replaceable>name</replaceable><literal>=</literal><replaceable>value</replaceable> that
681 specify options for the authentication method. Details about which
682 options are available for which authentication methods appear below.
683 </para>
684
685 <para>
686 In addition to the method-specific options listed below, there is a
687 method-independent authentication option <literal>clientcert</literal>, which
688 can be specified in any <literal>hostssl</literal> record.
689 This option can be set to <literal>verify-ca</literal> or
690 <literal>verify-full</literal>. Both options require the client
691 to present a valid (trusted) SSL certificate, while
692 <literal>verify-full</literal> additionally enforces that the
693 <literal>cn</literal> (Common Name) in the certificate matches
694 the username or an applicable mapping.
695 This behavior is similar to the <literal>cert</literal> authentication
696 method (see <xref linkend="auth-cert"/>) but enables pairing
697 the verification of client certificates with any authentication
698 method that supports <literal>hostssl</literal> entries.
699 </para>
700 <para>
701 On any record using client certificate authentication (i.e. one
702 using the <literal>cert</literal> authentication method or one
703 using the <literal>clientcert</literal> option), you can specify
704 which part of the client certificate credentials to match using
705 the <literal>clientname</literal> option. This option can have one
706 of two values. If you specify <literal>clientname=CN</literal>, which
707 is the default, the username is matched against the certificate's
708 <literal>Common Name (CN)</literal>. If instead you specify
709 <literal>clientname=DN</literal> the username is matched against the
710 entire <literal>Distinguished Name (DN)</literal> of the certificate.
711 This option is probably best used in conjunction with a username map.
712 The comparison is done with the <literal>DN</literal> in
713 <ulink url="https://datatracker.ietf.org/doc/html/rfc2253">RFC 2253</ulink>
714 format. To see the <literal>DN</literal> of a client certificate
715 in this format, do
716 <programlisting>
717 openssl x509 -in myclient.crt -noout -subject -nameopt RFC2253 | sed "s/^subject=//"
718 </programlisting>
719 Care needs to be taken when using this option, especially when using
720 regular expression matching against the <literal>DN</literal>.
721 </para>
722 </listitem>
723 </varlistentry>
724
725 <varlistentry>
726 <term><literal>include</literal></term>
727 <listitem>
728 <para>
729 This line will be replaced by the contents of the given file.
730 </para>
731 </listitem>
732 </varlistentry>
733
734 <varlistentry>
735 <term><literal>include_if_exists</literal></term>
736 <listitem>
737 <para>
738 This line will be replaced by the content of the given file if the
739 file exists. Otherwise, a message is logged to indicate that the file
740 has been skipped.
741 </para>
742 </listitem>
743 </varlistentry>
744
745 <varlistentry>
746 <term><literal>include_dir</literal></term>
747 <listitem>
748 <para>
749 This line will be replaced by the contents of all the files found in
750 the directory, if they don't start with a <literal>.</literal> and end
751 with <literal>.conf</literal>, processed in file name order (according
752 to C locale rules, i.e., numbers before letters, and uppercase letters
753 before lowercase ones).
754 </para>
755 </listitem>
756 </varlistentry>
757 </variablelist>
758 </para>
759
760 <para>
761 Files included by <literal>@</literal> constructs are read as lists of names,
762 which can be separated by either whitespace or commas. Comments are
763 introduced by <literal>#</literal>, just as in
764 <filename>pg_hba.conf</filename>, and nested <literal>@</literal> constructs are
765 allowed. Unless the file name following <literal>@</literal> is an absolute
766 path, it is taken to be relative to the directory containing the
767 referencing file.
768 </para>
769
770 <para>
771 Since the <filename>pg_hba.conf</filename> records are examined
772 sequentially for each connection attempt, the order of the records is
773 significant. Typically, earlier records will have tight connection
774 match parameters and weaker authentication methods, while later
775 records will have looser match parameters and stronger authentication
776 methods. For example, one might wish to use <literal>trust</literal>
777 authentication for local TCP/IP connections but require a password for
778 remote TCP/IP connections. In this case a record specifying
779 <literal>trust</literal> authentication for connections from 127.0.0.1 would
780 appear before a record specifying password authentication for a wider
781 range of allowed client IP addresses.
782 </para>
783
784 <tip>
785 <para>
786 To connect to a particular database, a user must not only pass the
787 <filename>pg_hba.conf</filename> checks, but must have the
788 <literal>CONNECT</literal> privilege for the database. If you wish to
789 restrict which users can connect to which databases, it's usually
790 easier to control this by granting/revoking <literal>CONNECT</literal> privilege
791 than to put the rules in <filename>pg_hba.conf</filename> entries.
792 </para>
793 </tip>
794
795 <para>
796 Some examples of <filename>pg_hba.conf</filename> entries are shown in
797 <xref linkend="example-pg-hba.conf"/>. See the next section for details on the
798 different authentication methods.
799 </para>
800
801 <example id="example-pg-hba.conf">
802 <title>Example <filename>pg_hba.conf</filename> Entries</title>
803 <programlisting>
804 # Allow any user on the local system to connect to any database with
805 # any database user name using Unix-domain sockets (the default for local
806 # connections).
807 #
808 # TYPE DATABASE USER ADDRESS METHOD
809 local all all trust
810
811 # The same using local loopback TCP/IP connections.
812 #
813 # TYPE DATABASE USER ADDRESS METHOD
814 host all all 127.0.0.1/32 trust
815
816 # The same as the previous line, but using a separate netmask column
817 #
818 # TYPE DATABASE USER IP-ADDRESS IP-MASK METHOD
819 host all all 127.0.0.1 255.255.255.255 trust
820
821 # The same over IPv6.
822 #
823 # TYPE DATABASE USER ADDRESS METHOD
824 host all all ::1/128 trust
825
826 # The same using a host name (would typically cover both IPv4 and IPv6).
827 #
828 # TYPE DATABASE USER ADDRESS METHOD
829 host all all localhost trust
830
831 # The same using a regular expression for DATABASE, that allows connection
832 # to any databases with a name beginning with "db" and finishing with a
833 # number using two to four digits (like "db1234" or "db12").
834 #
835 # TYPE DATABASE USER ADDRESS METHOD
836 host "/^db\d{2,4}$" all localhost trust
837
838 # Allow any user from any host with IP address 192.168.93.x to connect
839 # to database "postgres" as the same user name that ident reports for
840 # the connection (typically the operating system user name).
841 #
842 # TYPE DATABASE USER ADDRESS METHOD
843 host postgres all 192.168.93.0/24 ident
844
845 # Allow any user from host 192.168.12.10 to connect to database
846 # "postgres" if the user's password is correctly supplied.
847 #
848 # TYPE DATABASE USER ADDRESS METHOD
849 host postgres all 192.168.12.10/32 scram-sha-256
850
851 # Allow any user from hosts in the example.com domain to connect to
852 # any database if the user's password is correctly supplied.
853 #
854 # Require SCRAM authentication for most users, but make an exception
855 # for user 'mike', who uses an older client that doesn't support SCRAM
856 # authentication.
857 #
858 # TYPE DATABASE USER ADDRESS METHOD
859 host all mike .example.com md5
860 host all all .example.com scram-sha-256
861
862 # In the absence of preceding "host" lines, these three lines will
863 # reject all connections from 192.168.54.1 (since that entry will be
864 # matched first), but allow GSSAPI-encrypted connections from anywhere else
865 # on the Internet. The zero mask causes no bits of the host IP address to
866 # be considered, so it matches any host. Unencrypted GSSAPI connections
867 # (which "fall through" to the third line since "hostgssenc" only matches
868 # encrypted GSSAPI connections) are allowed, but only from 192.168.12.10.
869 #
870 # TYPE DATABASE USER ADDRESS METHOD
871 host all all 192.168.54.1/32 reject
872 hostgssenc all all 0.0.0.0/0 gss
873 host all all 192.168.12.10/32 gss
874
875 # Allow users from 192.168.x.x hosts to connect to any database, if
876 # they pass the ident check. If, for example, ident says the user is
877 # "bryanh" and he requests to connect as PostgreSQL user "guest1", the
878 # connection is allowed if there is an entry in pg_ident.conf for map
879 # "omicron" that says "bryanh" is allowed to connect as "guest1".
880 #
881 # TYPE DATABASE USER ADDRESS METHOD
882 host all all 192.168.0.0/16 ident map=omicron
883
884 # If these are the only four lines for local connections, they will
885 # allow local users to connect only to their own databases (databases
886 # with the same name as their database user name) except for users whose
887 # name end with "helpdesk", administrators and members of role "support",
888 # who can connect to all databases. The file $PGDATA/admins contains a
889 # list of names of administrators. Passwords are required in all cases.
890 #
891 # TYPE DATABASE USER ADDRESS METHOD
892 local sameuser all scram-sha-256
893 local all /^.*helpdesk$ scram-sha-256
894 local all @admins scram-sha-256
895 local all +support scram-sha-256
896
897 # The last two lines above can be combined into a single line:
898 local all @admins,+support scram-sha-256
899
900 # The database column can also use lists and file names:
901 local db1,db2,@demodbs all scram-sha-256
902 </programlisting>
903 </example>
904 </sect1>
905
906 <sect1 id="auth-username-maps">
907 <title>User Name Maps</title>
908
909 <indexterm zone="auth-username-maps">
910 <primary>User name maps</primary>
911 </indexterm>
912
913 <para>
914 When using an external authentication system such as Ident or GSSAPI,
915 the name of the operating system user that initiated the connection
916 might not be the same as the database user (role) that is to be used.
917 In this case, a user name map can be applied to map the operating system
918 user name to a database user. To use user name mapping, specify
919 <literal>map</literal>=<replaceable>map-name</replaceable>
920 in the options field in <filename>pg_hba.conf</filename>. This option is
921 supported for all authentication methods that receive external user names.
922 Since different mappings might be needed for different connections,
923 the name of the map to be used is specified in the
924 <replaceable>map-name</replaceable> parameter in <filename>pg_hba.conf</filename>
925 to indicate which map to use for each individual connection.
926 </para>
927
928 <para>
929 User name maps are defined in the ident map file, which by default is named
930 <filename>pg_ident.conf</filename><indexterm><primary>pg_ident.conf</primary></indexterm>
931 and is stored in the
932 cluster's data directory. (It is possible to place the map file
933 elsewhere, however; see the <xref linkend="guc-ident-file"/>
934 configuration parameter.)
935 The ident map file contains lines of the general forms:
936 <synopsis>
937 <replaceable>map-name</replaceable> <replaceable>system-username</replaceable> <replaceable>database-username</replaceable>
938 <replaceable>include</replaceable> <replaceable>file</replaceable>
939 <replaceable>include_if_exists</replaceable> <replaceable>file</replaceable>
940 <replaceable>include_dir</replaceable> <replaceable>directory</replaceable>
941 </synopsis>
942 Comments, whitespace and line continuations are handled in the same way as in
943 <filename>pg_hba.conf</filename>. The
944 <replaceable>map-name</replaceable> is an arbitrary name that will be used to
945 refer to this mapping in <filename>pg_hba.conf</filename>. The other
946 two fields specify an operating system user name and a matching
947 database user name. The same <replaceable>map-name</replaceable> can be
948 used repeatedly to specify multiple user-mappings within a single map.
949 </para>
950 <para>
951 As for <filename>pg_hba.conf</filename>, the lines in this file can
952 be include directives, following the same rules.
953 </para>
954
955 <para>
956 The <filename>pg_ident.conf</filename> file is read on start-up and
957 when the main server process receives a
958 <systemitem>SIGHUP</systemitem><indexterm><primary>SIGHUP</primary></indexterm>
959 signal. If you edit the file on an
960 active system, you will need to signal the postmaster
961 (using <literal>pg_ctl reload</literal>, calling the SQL function
962 <function>pg_reload_conf()</function>, or using <literal>kill
963 -HUP</literal>) to make it re-read the file.
964 </para>
965
966 <para>
967 The system view
968 <link linkend="view-pg-ident-file-mappings"><structname>pg_ident_file_mappings</structname></link>
969 can be helpful for pre-testing changes to the
970 <filename>pg_ident.conf</filename> file, or for diagnosing problems if
971 loading of the file did not have the desired effects. Rows in the view with
972 non-null <structfield>error</structfield> fields indicate problems in the
973 corresponding lines of the file.
974 </para>
975
976 <para>
977 There is no restriction regarding how many database users a given
978 operating system user can correspond to, nor vice versa. Thus, entries
979 in a map should be thought of as meaning <quote>this operating system
980 user is allowed to connect as this database user</quote>, rather than
981 implying that they are equivalent. The connection will be allowed if
982 there is any map entry that pairs the user name obtained from the
983 external authentication system with the database user name that the
984 user has requested to connect as. The value <literal>all</literal>
985 can be used as the <replaceable>database-username</replaceable> to specify
986 that if the <replaceable>system-username</replaceable> matches, then this
987 user is allowed to log in as any of the existing database users. Quoting
988 <literal>all</literal> makes the keyword lose its special meaning.
989 </para>
990 <para>
991 If the <replaceable>database-username</replaceable> begins with a
992 <literal>+</literal> character, then the operating system user can login as
993 any user belonging to that role, similarly to how user names beginning with
994 <literal>+</literal> are treated in <literal>pg_hba.conf</literal>.
995 Thus, a <literal>+</literal> mark means <quote>match any of the roles that
996 are directly or indirectly members of this role</quote>, while a name
997 without a <literal>+</literal> mark matches only that specific role. Quoting
998 a username starting with a <literal>+</literal> makes the
999 <literal>+</literal> lose its special meaning.
1000 </para>
1001 <para>
1002 If the <replaceable>system-username</replaceable> field starts with a slash (<literal>/</literal>),
1003 the remainder of the field is treated as a regular expression.
1004 (See <xref linkend="posix-syntax-details"/> for details of
1005 <productname>PostgreSQL</productname>'s regular expression syntax.) The regular
1006 expression can include a single capture, or parenthesized subexpression.
1007 The portion of the system user name that matched the capture can then
1008 be referenced in the <replaceable>database-username</replaceable>
1009 field as <literal>\1</literal> (backslash-one). This allows the mapping of
1010 multiple user names in a single line, which is particularly useful for
1011 simple syntax substitutions. For example, these entries
1012 <programlisting>
1013 mymap /^(.*)@mydomain\.com$ \1
1014 mymap /^(.*)@otherdomain\.com$ guest
1015 </programlisting>
1016 will remove the domain part for users with system user names that end with
1017 <literal>@mydomain.com</literal>, and allow any user whose system name ends with
1018 <literal>@otherdomain.com</literal> to log in as <literal>guest</literal>.
1019 Quoting a <replaceable>database-username</replaceable> containing
1020 <literal>\1</literal> <emphasis>does not</emphasis> make
1021 <literal>\1</literal> lose its special meaning.
1022 </para>
1023 <para>
1024 If the <replaceable>database-username</replaceable> field starts with
1025 a slash (<literal>/</literal>), the remainder of the field is treated
1026 as a regular expression.
1027 When the <replaceable>database-username</replaceable> field is a regular
1028 expression, it is not possible to use <literal>\1</literal> within it to
1029 refer to a capture from the <replaceable>system-username</replaceable>
1030 field.
1031 </para>
1032
1033 <tip>
1034 <para>
1035 Keep in mind that by default, a regular expression can match just part of
1036 a string. It's usually wise to use <literal>^</literal> and <literal>$</literal>, as
1037 shown in the above example, to force the match to be to the entire
1038 system user name.
1039 </para>
1040 </tip>
1041
1042 <para>
1043 A <filename>pg_ident.conf</filename> file that could be used in
1044 conjunction with the <filename>pg_hba.conf</filename> file in <xref
1045 linkend="example-pg-hba.conf"/> is shown in <xref
1046 linkend="example-pg-ident.conf"/>. In this example, anyone
1047 logged in to a machine on the 192.168 network that does not have the
1048 operating system user name <literal>bryanh</literal>, <literal>ann</literal>, or
1049 <literal>robert</literal> would not be granted access. Unix user
1050 <literal>robert</literal> would only be allowed access when he tries to
1051 connect as <productname>PostgreSQL</productname> user <literal>bob</literal>, not
1052 as <literal>robert</literal> or anyone else. <literal>ann</literal> would
1053 only be allowed to connect as <literal>ann</literal>. User
1054 <literal>bryanh</literal> would be allowed to connect as either
1055 <literal>bryanh</literal> or as <literal>guest1</literal>.
1056 </para>
1057
1058 <example id="example-pg-ident.conf">
1059 <title>An Example <filename>pg_ident.conf</filename> File</title>
1060 <programlisting>
1061 # MAPNAME SYSTEM-USERNAME PG-USERNAME
1062
1063 omicron bryanh bryanh
1064 omicron ann ann
1065 # bob has user name robert on these machines
1066 omicron robert bob
1067 # bryanh can also connect as guest1
1068 omicron bryanh guest1
1069 </programlisting>
1070 </example>
1071 </sect1>
1072
1073 <sect1 id="auth-methods">
1074 <title>Authentication Methods</title>
1075
1076 <para>
1077 <productname>PostgreSQL</productname> provides various methods for
1078 authenticating users:
1079
1080 <itemizedlist>
1081 <listitem>
1082 <para>
1083 <link linkend="auth-trust">Trust authentication</link>, which
1084 simply trusts that users are who they say they are.
1085 </para>
1086 </listitem>
1087 <listitem>
1088 <para>
1089 <link linkend="auth-password">Password authentication</link>, which
1090 requires that users send a password.
1091 </para>
1092 </listitem>
1093 <listitem>
1094 <para>
1095 <link linkend="gssapi-auth">GSSAPI authentication</link>, which
1096 relies on a GSSAPI-compatible security library. Typically this is
1097 used to access an authentication server such as a Kerberos or
1098 Microsoft Active Directory server.
1099 </para>
1100 </listitem>
1101 <listitem>
1102 <para>
1103 <link linkend="sspi-auth">SSPI authentication</link>, which
1104 uses a Windows-specific protocol similar to GSSAPI.
1105 </para>
1106 </listitem>
1107 <listitem>
1108 <para>
1109 <link linkend="auth-ident">Ident authentication</link>, which
1110 relies on an <quote>Identification Protocol</quote>
1111 (<ulink url="https://datatracker.ietf.org/doc/html/rfc1413">RFC 1413</ulink>)
1112 service on the client's machine. (On local Unix-socket connections,
1113 this is treated as peer authentication.)
1114 </para>
1115 </listitem>
1116 <listitem>
1117 <para>
1118 <link linkend="auth-peer">Peer authentication</link>, which
1119 relies on operating system facilities to identify the process at the
1120 other end of a local connection. This is not supported for remote
1121 connections.
1122 </para>
1123 </listitem>
1124 <listitem>
1125 <para>
1126 <link linkend="auth-ldap">LDAP authentication</link>, which
1127 relies on an LDAP authentication server.
1128 </para>
1129 </listitem>
1130 <listitem>
1131 <para>
1132 <link linkend="auth-radius">RADIUS authentication</link>, which
1133 relies on a RADIUS authentication server.
1134 </para>
1135 </listitem>
1136 <listitem>
1137 <para>
1138 <link linkend="auth-cert">Certificate authentication</link>, which
1139 requires an SSL connection and authenticates users by checking the
1140 SSL certificate they send.
1141 </para>
1142 </listitem>
1143 <listitem>
1144 <para>
1145 <link linkend="auth-pam">PAM authentication</link>, which
1146 relies on a PAM (Pluggable Authentication Modules) library.
1147 </para>
1148 </listitem>
1149 <listitem>
1150 <para>
1151 <link linkend="auth-bsd">BSD authentication</link>, which
1152 relies on the BSD Authentication framework (currently available
1153 only on OpenBSD).
1154 </para>
1155 </listitem>
1156 <listitem>
1157 <para>
1158 <link linkend="auth-oauth">OAuth authorization/authentication</link>,
1159 which relies on an external OAuth 2.0 identity provider.
1160 </para>
1161 </listitem>
1162 </itemizedlist>
1163 </para>
1164
1165 <para>
1166 Peer authentication is usually recommendable for local connections,
1167 though trust authentication might be sufficient in some circumstances.
1168 Password authentication is the easiest choice for remote connections.
1169 All the other options require some kind of external security
1170 infrastructure (usually an authentication server or a certificate
1171 authority for issuing SSL certificates), or are platform-specific.
1172 </para>
1173
1174 <para>
1175 The following sections describe each of these authentication methods
1176 in more detail.
1177 </para>
1178 </sect1>
1179
1180 <sect1 id="auth-trust">
1181 <title>Trust Authentication</title>
1182
1183 <para>
1184 When <literal>trust</literal> authentication is specified,
1185 <productname>PostgreSQL</productname> assumes that anyone who can
1186 connect to the server is authorized to access the database with
1187 whatever database user name they specify (even superuser names).
1188 Of course, restrictions made in the <literal>database</literal> and
1189 <literal>user</literal> columns still apply.
1190 This method should only be used when there is adequate
1191 operating-system-level protection on connections to the server.
1192 </para>
1193
1194 <para>
1195 <literal>trust</literal> authentication is appropriate and very
1196 convenient for local connections on a single-user workstation. It
1197 is usually <emphasis>not</emphasis> appropriate by itself on a multiuser
1198 machine. However, you might be able to use <literal>trust</literal> even
1199 on a multiuser machine, if you restrict access to the server's
1200 Unix-domain socket file using file-system permissions. To do this, set the
1201 <varname>unix_socket_permissions</varname> (and possibly
1202 <varname>unix_socket_group</varname>) configuration parameters as
1203 described in <xref linkend="runtime-config-connection"/>. Or you
1204 could set the <varname>unix_socket_directories</varname>
1205 configuration parameter to place the socket file in a suitably
1206 restricted directory.
1207 </para>
1208
1209 <para>
1210 Setting file-system permissions only helps for Unix-socket connections.
1211 Local TCP/IP connections are not restricted by file-system permissions.
1212 Therefore, if you want to use file-system permissions for local security,
1213 remove the <literal>host ... 127.0.0.1 ...</literal> line from
1214 <filename>pg_hba.conf</filename>, or change it to a
1215 non-<literal>trust</literal> authentication method.
1216 </para>
1217
1218 <para>
1219 <literal>trust</literal> authentication is only suitable for TCP/IP connections
1220 if you trust every user on every machine that is allowed to connect
1221 to the server by the <filename>pg_hba.conf</filename> lines that specify
1222 <literal>trust</literal>. It is seldom reasonable to use <literal>trust</literal>
1223 for any TCP/IP connections other than those from <systemitem>localhost</systemitem> (127.0.0.1).
1224 </para>
1225
1226 </sect1>
1227
1228 <sect1 id="auth-password">
1229 <title>Password Authentication</title>
1230
1231 <indexterm>
1232 <primary>MD5</primary>
1233 </indexterm>
1234 <indexterm>
1235 <primary>SCRAM</primary>
1236 </indexterm>
1237 <indexterm>
1238 <primary>password</primary>
1239 <secondary>authentication</secondary>
1240 </indexterm>
1241
1242 <para>
1243 There are several password-based authentication methods. These methods
1244 operate similarly but differ in how the users' passwords are stored on the
1245 server and how the password provided by a client is sent across the
1246 connection.
1247 </para>
1248
1249 <variablelist>
1250 <varlistentry>
1251 <term><literal>scram-sha-256</literal></term>
1252 <listitem>
1253 <para>
1254 The method <literal>scram-sha-256</literal> performs SCRAM-SHA-256
1255 authentication, as described in
1256 <ulink url="https://datatracker.ietf.org/doc/html/rfc7677">RFC 7677</ulink>. It
1257 is a challenge-response scheme that prevents password sniffing on
1258 untrusted connections and supports storing passwords on the server in a
1259 cryptographically hashed form that is thought to be secure.
1260 </para>
1261
1262 <para>
1263 This is the most secure of the currently provided methods, but it is
1264 not supported by older client libraries.
1265 </para>
1266 </listitem>
1267 </varlistentry>
1268
1269 <varlistentry>
1270 <term><literal>md5</literal></term>
1271 <listitem>
1272 <para>
1273 The method <literal>md5</literal> uses a custom less secure challenge-response
1274 mechanism. It prevents password sniffing and avoids storing passwords
1275 on the server in plain text but provides no protection if an attacker
1276 manages to steal the password hash from the server. Also, the MD5 hash
1277 algorithm is nowadays no longer considered secure against determined
1278 attacks.
1279 </para>
1280
1281 <para>
1282 To ease transition from the <literal>md5</literal> method to the newer
1283 SCRAM method, if <literal>md5</literal> is specified as a method
1284 in <filename>pg_hba.conf</filename> but the user's password on the
1285 server is encrypted for SCRAM (see below), then SCRAM-based
1286 authentication will automatically be chosen instead.
1287 </para>
1288
1289 <warning>
1290 <para>
1291 Support for MD5-encrypted passwords is deprecated and will be removed
1292 in a future release of <productname>PostgreSQL</productname>. Refer to
1293 the text below for details about migrating to another password type.
1294 </para>
1295 </warning>
1296 </listitem>
1297 </varlistentry>
1298
1299 <varlistentry>
1300 <term><literal>password</literal></term>
1301 <listitem>
1302 <para>
1303 The method <literal>password</literal> sends the password in clear-text and is
1304 therefore vulnerable to password <quote>sniffing</quote> attacks. It should
1305 always be avoided if possible. If the connection is protected by SSL
1306 encryption then <literal>password</literal> can be used safely, though.
1307 (Though SSL certificate authentication might be a better choice if one
1308 is depending on using SSL).
1309 </para>
1310 </listitem>
1311 </varlistentry>
1312 </variablelist>
1313
1314 <para>
1315 <productname>PostgreSQL</productname> database passwords are
1316 separate from operating system user passwords. The password for
1317 each database user is stored in the <literal>pg_authid</literal> system
1318 catalog. Passwords can be managed with the SQL commands
1319 <xref linkend="sql-createrole"/> and
1320 <xref linkend="sql-alterrole"/>,
1321 e.g., <userinput>CREATE ROLE foo WITH LOGIN PASSWORD 'secret'</userinput>,
1322 or the <application>psql</application>
1323 command <literal>\password</literal>.
1324 If no password has been set up for a user, the stored password
1325 is null and password authentication will always fail for that user.
1326 </para>
1327
1328 <para>
1329 The availability of the different password-based authentication methods
1330 depends on how a user's password on the server is encrypted (or hashed,
1331 more accurately). This is controlled by the configuration
1332 parameter <xref linkend="guc-password-encryption"/> at the time the
1333 password is set. If a password was encrypted using
1334 the <literal>scram-sha-256</literal> setting, then it can be used for the
1335 authentication methods <literal>scram-sha-256</literal>
1336 and <literal>password</literal> (but password transmission will be in
1337 plain text in the latter case). The authentication method
1338 specification <literal>md5</literal> will automatically switch to using
1339 the <literal>scram-sha-256</literal> method in this case, as explained
1340 above, so it will also work. If a password was encrypted using
1341 the <literal>md5</literal> setting, then it can be used only for
1342 the <literal>md5</literal> and <literal>password</literal> authentication
1343 method specifications (again, with the password transmitted in plain text
1344 in the latter case). (Previous PostgreSQL releases supported storing the
1345 password on the server in plain text. This is no longer possible.) To
1346 check the currently stored password hashes, see the system
1347 catalog <literal>pg_authid</literal>.
1348 </para>
1349
1350 <para>
1351 To upgrade an existing installation from <literal>md5</literal>
1352 to <literal>scram-sha-256</literal>, after having ensured that all client
1353 libraries in use are new enough to support SCRAM,
1354 set <literal>password_encryption = 'scram-sha-256'</literal>
1355 in <filename>postgresql.conf</filename>, make all users set new passwords,
1356 and change the authentication method specifications
1357 in <filename>pg_hba.conf</filename> to <literal>scram-sha-256</literal>.
1358 </para>
1359 </sect1>
1360
1361 <sect1 id="gssapi-auth">
1362 <title>GSSAPI Authentication</title>
1363
1364 <indexterm zone="gssapi-auth">
1365 <primary>GSSAPI</primary>
1366 </indexterm>
1367
1368 <para>
1369 <productname>GSSAPI</productname> is an industry-standard protocol
1370 for secure authentication defined in
1371 <ulink url="https://datatracker.ietf.org/doc/html/rfc2743">RFC 2743</ulink>.
1372 <productname>PostgreSQL</productname>
1373 supports <productname>GSSAPI</productname> for authentication,
1374 communications encryption, or both.
1375 <productname>GSSAPI</productname> provides automatic authentication
1376 (single sign-on) for systems that support it. The authentication itself is
1377 secure. If <productname>GSSAPI</productname> encryption
1378 or <acronym>SSL</acronym> encryption is
1379 used, the data sent along the database connection will be encrypted;
1380 otherwise, it will not.
1381 </para>
1382
1383 <para>
1384 GSSAPI support has to be enabled when <productname>PostgreSQL</productname> is built;
1385 see <xref linkend="installation"/> for more information.
1386 </para>
1387
1388 <para>
1389 When <productname>GSSAPI</productname> uses
1390 <productname>Kerberos</productname>, it uses a standard service
1391 principal (authentication identity) name in the format
1392 <literal><replaceable>servicename</replaceable>/<replaceable>hostname</replaceable>@<replaceable>realm</replaceable></literal>.
1393 The principal name used by a particular installation is not encoded in
1394 the <productname>PostgreSQL</productname> server in any way; rather it
1395 is specified in the <firstterm>keytab</firstterm> file that the server
1396 reads to determine its identity. If multiple principals are listed in
1397 the keytab file, the server will accept any one of them.
1398 The server's realm name is the preferred realm specified in the Kerberos
1399 configuration file(s) accessible to the server.
1400 </para>
1401
1402 <para>
1403 When connecting, the client must know the principal name of the server
1404 it intends to connect to. The <replaceable>servicename</replaceable>
1405 part of the principal is ordinarily <literal>postgres</literal>,
1406 but another value can be selected via <application>libpq</application>'s
1407 <xref linkend="libpq-connect-krbsrvname"/> connection parameter.
1408 The <replaceable>hostname</replaceable> part is the fully qualified
1409 host name that <application>libpq</application> is told to connect to.
1410 The realm name is the preferred realm specified in the Kerberos
1411 configuration file(s) accessible to the client.
1412 </para>
1413
1414 <para>
1415 The client will also have a principal name for its own identity
1416 (and it must have a valid ticket for this principal). To
1417 use <productname>GSSAPI</productname> for authentication, the client
1418 principal must be associated with
1419 a <productname>PostgreSQL</productname> database user name.
1420 The <filename>pg_ident.conf</filename> configuration file can be used
1421 to map principals to user names; for example,
1422 <literal>pgusername@realm</literal> could be mapped to just <literal>pgusername</literal>.
1423 Alternatively, you can use the full <literal>username@realm</literal> principal as
1424 the role name in <productname>PostgreSQL</productname> without any mapping.
1425 </para>
1426
1427 <para>
1428 <productname>PostgreSQL</productname> also supports mapping
1429 client principals to user names by just stripping the realm from
1430 the principal. This method is supported for backwards compatibility and is
1431 strongly discouraged as it is then impossible to distinguish different users
1432 with the same user name but coming from different realms. To enable this,
1433 set <literal>include_realm</literal> to 0. For simple single-realm
1434 installations, doing that combined with setting the
1435 <literal>krb_realm</literal> parameter (which checks that the principal's realm
1436 matches exactly what is in the <literal>krb_realm</literal> parameter)
1437 is still secure; but this is a
1438 less capable approach compared to specifying an explicit mapping in
1439 <filename>pg_ident.conf</filename>.
1440 </para>
1441
1442 <para>
1443 The location of the server's keytab file is specified by the <xref
1444 linkend="guc-krb-server-keyfile"/> configuration parameter.
1445 For security reasons, it is recommended to use a separate keytab
1446 just for the <productname>PostgreSQL</productname> server rather
1447 than allowing the server to read the system keytab file.
1448 Make sure that your server keytab file is readable (and preferably
1449 only readable, not writable) by the <productname>PostgreSQL</productname>
1450 server account. (See also <xref linkend="postgres-user"/>.)
1451 </para>
1452
1453 <para>
1454 The keytab file is generated using the Kerberos software; see the
1455 Kerberos documentation for details. The following example shows
1456 doing this using the <application>kadmin</application> tool of
1457 MIT Kerberos:
1458 <screen>
1459 <prompt>kadmin% </prompt><userinput>addprinc -randkey postgres/server.my.domain.org</userinput>
1460 <prompt>kadmin% </prompt><userinput>ktadd -k krb5.keytab postgres/server.my.domain.org</userinput>
1461 </screen>
1462 </para>
1463
1464 <para>
1465 The following authentication options are supported for
1466 the <productname>GSSAPI</productname> authentication method:
1467 <variablelist>
1468 <varlistentry>
1469 <term><literal>include_realm</literal></term>
1470 <listitem>
1471 <para>
1472 If set to 0, the realm name from the authenticated user principal is
1473 stripped off before being passed through the user name mapping
1474 (<xref linkend="auth-username-maps"/>). This is discouraged and is
1475 primarily available for backwards compatibility, as it is not secure
1476 in multi-realm environments unless <literal>krb_realm</literal> is
1477 also used. It is recommended to
1478 leave <literal>include_realm</literal> set to the default (1) and to
1479 provide an explicit mapping in <filename>pg_ident.conf</filename> to convert
1480 principal names to <productname>PostgreSQL</productname> user names.
1481 </para>
1482 </listitem>
1483 </varlistentry>
1484
1485 <varlistentry>
1486 <term><literal>map</literal></term>
1487 <listitem>
1488 <para>
1489 Allows mapping from client principals to database user names. See
1490 <xref linkend="auth-username-maps"/> for details. For a GSSAPI/Kerberos
1491 principal, such as <literal>[email protected]</literal> (or, less
1492 commonly, <literal>username/[email protected]</literal>), the
1493 user name used for mapping is
1494 <literal>[email protected]</literal> (or
1495 <literal>username/[email protected]</literal>, respectively),
1496 unless <literal>include_realm</literal> has been set to 0, in which case
1497 <literal>username</literal> (or <literal>username/hostbased</literal>)
1498 is what is seen as the system user name when mapping.
1499 </para>
1500 </listitem>
1501 </varlistentry>
1502
1503 <varlistentry>
1504 <term><literal>krb_realm</literal></term>
1505 <listitem>
1506 <para>
1507 Sets the realm to match user principal names against. If this parameter
1508 is set, only users of that realm will be accepted. If it is not set,
1509 users of any realm can connect, subject to whatever user name mapping
1510 is done.
1511 </para>
1512 </listitem>
1513 </varlistentry>
1514 </variablelist>
1515 </para>
1516
1517 <para>
1518 In addition to these settings, which can be different for
1519 different <filename>pg_hba.conf</filename> entries, there is the
1520 server-wide <xref linkend="guc-krb-caseins-users"/> configuration
1521 parameter. If that is set to true, client principals are matched to
1522 user map entries case-insensitively. <literal>krb_realm</literal>, if
1523 set, is also matched case-insensitively.
1524 </para>
1525 </sect1>
1526
1527 <sect1 id="sspi-auth">
1528 <title>SSPI Authentication</title>
1529
1530 <indexterm zone="sspi-auth">
1531 <primary>SSPI</primary>
1532 </indexterm>
1533
1534 <para>
1535 <productname>SSPI</productname> is a <productname>Windows</productname>
1536 technology for secure authentication with single sign-on.
1537 <productname>PostgreSQL</productname> will use SSPI in
1538 <literal>negotiate</literal> mode, which will use
1539 <productname>Kerberos</productname> when possible and automatically
1540 fall back to <productname>NTLM</productname> in other cases.
1541 <productname>SSPI</productname> and <productname>GSSAPI</productname>
1542 interoperate as clients and servers, e.g., an
1543 <productname>SSPI</productname> client can authenticate to an
1544 <productname>GSSAPI</productname> server. It is recommended to use
1545 <productname>SSPI</productname> on Windows clients and servers and
1546 <productname>GSSAPI</productname> on non-Windows platforms.
1547 </para>
1548
1549 <para>
1550 When using <productname>Kerberos</productname> authentication,
1551 <productname>SSPI</productname> works the same way
1552 <productname>GSSAPI</productname> does; see <xref linkend="gssapi-auth"/>
1553 for details.
1554 </para>
1555
1556 <para>
1557 The following configuration options are supported for <productname>SSPI</productname>:
1558 <variablelist>
1559
1560 <varlistentry>
1561 <term><literal>include_realm</literal></term>
1562 <listitem>
1563 <para>
1564 If set to 0, the realm name from the authenticated user principal is
1565 stripped off before being passed through the user name mapping
1566 (<xref linkend="auth-username-maps"/>). This is discouraged and is
1567 primarily available for backwards compatibility, as it is not secure
1568 in multi-realm environments unless <literal>krb_realm</literal> is
1569 also used. It is recommended to
1570 leave <literal>include_realm</literal> set to the default (1) and to
1571 provide an explicit mapping in <filename>pg_ident.conf</filename> to convert
1572 principal names to <productname>PostgreSQL</productname> user names.
1573 </para>
1574 </listitem>
1575 </varlistentry>
1576
1577 <varlistentry>
1578 <term><literal>compat_realm</literal></term>
1579 <listitem>
1580 <para>
1581 If set to 1, the domain's SAM-compatible name (also known as the
1582 NetBIOS name) is used for the <literal>include_realm</literal>
1583 option. This is the default. If set to 0, the true realm name from
1584 the Kerberos user principal name is used.
1585 </para>
1586 <para>
1587 Do not disable this option unless your server runs under a domain
1588 account (this includes virtual service accounts on a domain member
1589 system) and all clients authenticating through SSPI are also using
1590 domain accounts, or authentication will fail.
1591 </para>
1592 </listitem>
1593 </varlistentry>
1594
1595 <varlistentry>
1596 <term><literal>upn_username</literal></term>
1597 <listitem>
1598 <para>
1599 If this option is enabled along with <literal>compat_realm</literal>,
1600 the user name from the Kerberos UPN is used for authentication. If
1601 it is disabled (the default), the SAM-compatible user name is used.
1602 By default, these two names are identical for new user accounts.
1603 </para>
1604 <para>
1605 Note that <application>libpq</application> uses the SAM-compatible name if no
1606 explicit user name is specified. If you use
1607 <application>libpq</application> or a driver based on it, you should
1608 leave this option disabled or explicitly specify user name in the
1609 connection string.
1610 </para>
1611 </listitem>
1612 </varlistentry>
1613
1614 <varlistentry>
1615 <term><literal>map</literal></term>
1616 <listitem>
1617 <para>
1618 Allows for mapping between system and database user names. See
1619 <xref linkend="auth-username-maps"/> for details. For an SSPI/Kerberos
1620 principal, such as <literal>[email protected]</literal> (or, less
1621 commonly, <literal>username/[email protected]</literal>), the
1622 user name used for mapping is
1623 <literal>[email protected]</literal> (or
1624 <literal>username/[email protected]</literal>, respectively),
1625 unless <literal>include_realm</literal> has been set to 0, in which case
1626 <literal>username</literal> (or <literal>username/hostbased</literal>)
1627 is what is seen as the system user name when mapping.
1628 </para>
1629 </listitem>
1630 </varlistentry>
1631
1632 <varlistentry>
1633 <term><literal>krb_realm</literal></term>
1634 <listitem>
1635 <para>
1636 Sets the realm to match user principal names against. If this parameter
1637 is set, only users of that realm will be accepted. If it is not set,
1638 users of any realm can connect, subject to whatever user name mapping
1639 is done.
1640 </para>
1641 </listitem>
1642 </varlistentry>
1643 </variablelist>
1644 </para>
1645 </sect1>
1646
1647 <sect1 id="auth-ident">
1648 <title>Ident Authentication</title>
1649
1650 <indexterm>
1651 <primary>ident</primary>
1652 </indexterm>
1653
1654 <para>
1655 The ident authentication method works by obtaining the client's
1656 operating system user name from an ident server and using it as
1657 the allowed database user name (with an optional user name mapping).
1658 This is only supported on TCP/IP connections.
1659 </para>
1660
1661 <note>
1662 <para>
1663 When ident is specified for a local (non-TCP/IP) connection,
1664 peer authentication (see <xref linkend="auth-peer"/>) will be
1665 used instead.
1666 </para>
1667 </note>
1668
1669 <para>
1670 The following configuration options are supported for <literal>ident</literal>:
1671 <variablelist>
1672 <varlistentry>
1673 <term><literal>map</literal></term>
1674 <listitem>
1675 <para>
1676 Allows for mapping between system and database user names. See
1677 <xref linkend="auth-username-maps"/> for details.
1678 </para>
1679 </listitem>
1680 </varlistentry>
1681 </variablelist>
1682 </para>
1683
1684 <para>
1685 The <quote>Identification Protocol</quote> is described in
1686 <ulink url="https://datatracker.ietf.org/doc/html/rfc1413">RFC 1413</ulink>.
1687 Virtually every Unix-like
1688 operating system ships with an ident server that listens on TCP
1689 port 113 by default. The basic functionality of an ident server
1690 is to answer questions like <quote>What user initiated the
1691 connection that goes out of your port <replaceable>X</replaceable>
1692 and connects to my port <replaceable>Y</replaceable>?</quote>.
1693 Since <productname>PostgreSQL</productname> knows both <replaceable>X</replaceable> and
1694 <replaceable>Y</replaceable> when a physical connection is established, it
1695 can interrogate the ident server on the host of the connecting
1696 client and can theoretically determine the operating system user
1697 for any given connection.
1698 </para>
1699
1700 <para>
1701 The drawback of this procedure is that it depends on the integrity
1702 of the client: if the client machine is untrusted or compromised,
1703 an attacker could run just about any program on port 113 and
1704 return any user name they choose. This authentication method is
1705 therefore only appropriate for closed networks where each client
1706 machine is under tight control and where the database and system
1707 administrators operate in close contact. In other words, you must
1708 trust the machine running the ident server.
1709 Heed the warning:
1710 <blockquote>
1711 <attribution>RFC 1413</attribution>
1712 <para>
1713 The Identification Protocol is not intended as an authorization
1714 or access control protocol.
1715 </para>
1716 </blockquote>
1717 </para>
1718
1719 <para>
1720 Some ident servers have a nonstandard option that causes the returned
1721 user name to be encrypted, using a key that only the originating
1722 machine's administrator knows. This option <emphasis>must not</emphasis> be
1723 used when using the ident server with <productname>PostgreSQL</productname>,
1724 since <productname>PostgreSQL</productname> does not have any way to decrypt the
1725 returned string to determine the actual user name.
1726 </para>
1727 </sect1>
1728
1729 <sect1 id="auth-peer">
1730 <title>Peer Authentication</title>
1731
1732 <indexterm>
1733 <primary>peer</primary>
1734 </indexterm>
1735
1736 <para>
1737 The peer authentication method works by obtaining the client's
1738 operating system user name from the kernel and using it as the
1739 allowed database user name (with optional user name mapping). This
1740 method is only supported on local connections.
1741 </para>
1742
1743 <para>
1744 The following configuration options are supported for <literal>peer</literal>:
1745 <variablelist>
1746 <varlistentry>
1747 <term><literal>map</literal></term>
1748 <listitem>
1749 <para>
1750 Allows for mapping between system and database user names. See
1751 <xref linkend="auth-username-maps"/> for details.
1752 </para>
1753 </listitem>
1754 </varlistentry>
1755 </variablelist>
1756 </para>
1757
1758 <para>
1759 Peer authentication is only available on operating systems providing
1760 the <function>getpeereid()</function> function, the <symbol>SO_PEERCRED</symbol>
1761 socket parameter, or similar mechanisms. Currently that includes
1762 <systemitem class="osname">Linux</systemitem>,
1763 most flavors of <systemitem class="osname">BSD</systemitem> including
1764 <systemitem class="osname">macOS</systemitem>,
1765 and <systemitem class="osname">Solaris</systemitem>.
1766 </para>
1767
1768 </sect1>
1769
1770 <sect1 id="auth-ldap">
1771 <title>LDAP Authentication</title>
1772
1773 <indexterm zone="auth-ldap">
1774 <primary>LDAP</primary>
1775 </indexterm>
1776
1777 <para>
1778 This authentication method operates similarly to
1779 <literal>password</literal> except that it uses LDAP
1780 as the password verification method. LDAP is used only to validate
1781 the user name/password pairs. Therefore the user must already
1782 exist in the database before LDAP can be used for
1783 authentication.
1784 </para>
1785
1786 <para>
1787 LDAP authentication can operate in two modes. In the first mode,
1788 which we will call the simple bind mode,
1789 the server will bind to the distinguished name constructed as
1790 <replaceable>prefix</replaceable> <replaceable>username</replaceable> <replaceable>suffix</replaceable>.
1791 Typically, the <replaceable>prefix</replaceable> parameter is used to specify
1792 <literal>cn=</literal>, or <replaceable>DOMAIN</replaceable><literal>\</literal> in an Active
1793 Directory environment. <replaceable>suffix</replaceable> is used to specify the
1794 remaining part of the DN in a non-Active Directory environment.
1795 </para>
1796
1797 <para>
1798 In the second mode, which we will call the search+bind mode,
1799 the server first binds to the LDAP directory with
1800 a fixed user name and password, specified with <replaceable>ldapbinddn</replaceable>
1801 and <replaceable>ldapbindpasswd</replaceable>, and performs a search for the user trying
1802 to log in to the database. If no user and password is configured, an
1803 anonymous bind will be attempted to the directory. The search will be
1804 performed over the subtree at <replaceable>ldapbasedn</replaceable>, and will try to
1805 do an exact match of the attribute specified in
1806 <replaceable>ldapsearchattribute</replaceable>.
1807 Once the user has been found in
1808 this search, the server re-binds to the directory as
1809 this user, using the password specified by the client, to verify that the
1810 login is correct. This mode is the same as that used by LDAP authentication
1811 schemes in other software, such as Apache <literal>mod_authnz_ldap</literal> and <literal>pam_ldap</literal>.
1812 This method allows for significantly more flexibility
1813 in where the user objects are located in the directory, but will cause
1814 two additional requests to the LDAP server to be made.
1815 </para>
1816
1817 <para>
1818 The following configuration options are used in both modes:
1819 <variablelist>
1820 <varlistentry>
1821 <term><literal>ldapserver</literal></term>
1822 <listitem>
1823 <para>
1824 Names or IP addresses of LDAP servers to connect to. Multiple
1825 servers may be specified, separated by spaces.
1826 </para>
1827 </listitem>
1828 </varlistentry>
1829 <varlistentry>
1830 <term><literal>ldapport</literal></term>
1831 <listitem>
1832 <para>
1833 Port number on LDAP server to connect to. If no port is specified,
1834 the LDAP library's default port setting will be used.
1835 </para>
1836 </listitem>
1837 </varlistentry>
1838 <varlistentry>
1839 <term><literal>ldapscheme</literal></term>
1840 <listitem>
1841 <para>
1842 Set to <literal>ldaps</literal> to use LDAPS. This is a non-standard
1843 way of using LDAP over SSL, supported by some LDAP server
1844 implementations. See also the <literal>ldaptls</literal> option for
1845 an alternative.
1846 </para>
1847 </listitem>
1848 </varlistentry>
1849 <varlistentry>
1850 <term><literal>ldaptls</literal></term>
1851 <listitem>
1852 <para>
1853 Set to 1 to make the connection between PostgreSQL and the LDAP server
1854 use TLS encryption. This uses the <literal>StartTLS</literal>
1855 operation per <ulink url="https://datatracker.ietf.org/doc/html/rfc4513">RFC 4513</ulink>.
1856 See also the <literal>ldapscheme</literal> option for an alternative.
1857 </para>
1858 </listitem>
1859 </varlistentry>
1860 </variablelist>
1861 </para>
1862
1863 <para>
1864 Note that using <literal>ldapscheme</literal> or
1865 <literal>ldaptls</literal> only encrypts the traffic between the
1866 PostgreSQL server and the LDAP server. The connection between the
1867 PostgreSQL server and the PostgreSQL client will still be unencrypted
1868 unless SSL is used there as well.
1869 </para>
1870
1871 <para>
1872 The following options are used in simple bind mode only:
1873 <variablelist>
1874 <varlistentry>
1875 <term><literal>ldapprefix</literal></term>
1876 <listitem>
1877 <para>
1878 String to prepend to the user name when forming the DN to bind as,
1879 when doing simple bind authentication.
1880 </para>
1881 </listitem>
1882 </varlistentry>
1883 <varlistentry>
1884 <term><literal>ldapsuffix</literal></term>
1885 <listitem>
1886 <para>
1887 String to append to the user name when forming the DN to bind as,
1888 when doing simple bind authentication.
1889 </para>
1890 </listitem>
1891 </varlistentry>
1892 </variablelist>
1893 </para>
1894
1895 <para>
1896 The following options are used in search+bind mode only:
1897 <variablelist>
1898 <varlistentry>
1899 <term><literal>ldapbasedn</literal></term>
1900 <listitem>
1901 <para>
1902 Root DN to begin the search for the user in, when doing search+bind
1903 authentication.
1904 </para>
1905 </listitem>
1906 </varlistentry>
1907 <varlistentry>
1908 <term><literal>ldapbinddn</literal></term>
1909 <listitem>
1910 <para>
1911 DN of user to bind to the directory with to perform the search when
1912 doing search+bind authentication.
1913 </para>
1914 </listitem>
1915 </varlistentry>
1916 <varlistentry>
1917 <term><literal>ldapbindpasswd</literal></term>
1918 <listitem>
1919 <para>
1920 Password for user to bind to the directory with to perform the search
1921 when doing search+bind authentication.
1922 </para>
1923 </listitem>
1924 </varlistentry>
1925 <varlistentry>
1926 <term><literal>ldapsearchattribute</literal></term>
1927 <listitem>
1928 <para>
1929 Attribute to match against the user name in the search when doing
1930 search+bind authentication. If no attribute is specified, the
1931 <literal>uid</literal> attribute will be used.
1932 </para>
1933 </listitem>
1934 </varlistentry>
1935 <varlistentry>
1936 <term><literal>ldapsearchfilter</literal></term>
1937 <listitem>
1938 <para>
1939 The search filter to use when doing search+bind authentication.
1940 Occurrences of <literal>$username</literal> will be replaced with the
1941 user name. This allows for more flexible search filters than
1942 <literal>ldapsearchattribute</literal>.
1943 </para>
1944 </listitem>
1945 </varlistentry>
1946 </variablelist>
1947 </para>
1948
1949 <para>
1950 The following option may be used as an alternative way to write some of the
1951 above LDAP options in a more compact and standard form:
1952 <variablelist>
1953 <varlistentry>
1954 <term><literal>ldapurl</literal></term>
1955 <listitem>
1956 <para>
1957 An <ulink url="https://datatracker.ietf.org/doc/html/rfc4516">RFC 4516</ulink>
1958 LDAP URL. The format is
1959 <synopsis>
1960 ldap[s]://<replaceable>host</replaceable>[:<replaceable>port</replaceable>]/<replaceable>basedn</replaceable>[?[<replaceable>attribute</replaceable>][?[<replaceable>scope</replaceable>][?[<replaceable>filter</replaceable>]]]]
1961 </synopsis>
1962 <replaceable>scope</replaceable> must be one
1963 of <literal>base</literal>, <literal>one</literal>, <literal>sub</literal>,
1964 typically the last. (The default is <literal>base</literal>, which
1965 is normally not useful in this application.) <replaceable>attribute</replaceable> can
1966 nominate a single attribute, in which case it is used as a value for
1967 <literal>ldapsearchattribute</literal>. If
1968 <replaceable>attribute</replaceable> is empty then
1969 <replaceable>filter</replaceable> can be used as a value for
1970 <literal>ldapsearchfilter</literal>.
1971 </para>
1972
1973 <para>
1974 The URL scheme <literal>ldaps</literal> chooses the LDAPS method for
1975 making LDAP connections over SSL, equivalent to using
1976 <literal>ldapscheme=ldaps</literal>. To use encrypted LDAP
1977 connections using the <literal>StartTLS</literal> operation, use the
1978 normal URL scheme <literal>ldap</literal> and specify the
1979 <literal>ldaptls</literal> option in addition to
1980 <literal>ldapurl</literal>.
1981 </para>
1982
1983 <para>
1984 For non-anonymous binds, <literal>ldapbinddn</literal>
1985 and <literal>ldapbindpasswd</literal> must be specified as separate
1986 options.
1987 </para>
1988
1989 <para>
1990 LDAP URLs are currently only supported with
1991 <productname>OpenLDAP</productname>, not on Windows.
1992 </para>
1993 </listitem>
1994 </varlistentry>
1995 </variablelist>
1996 </para>
1997
1998 <para>
1999 It is an error to mix configuration options for simple bind with options
2000 for search+bind. To use <literal>ldapurl</literal> in simple bind mode, the
2001 URL must not contain a <literal>basedn</literal> or query elements.
2002 </para>
2003
2004 <para>
2005 When using search+bind mode, the search can be performed using a single
2006 attribute specified with <literal>ldapsearchattribute</literal>, or using
2007 a custom search filter specified with
2008 <literal>ldapsearchfilter</literal>.
2009 Specifying <literal>ldapsearchattribute=foo</literal> is equivalent to
2010 specifying <literal>ldapsearchfilter="(foo=$username)"</literal>. If neither
2011 option is specified the default is
2012 <literal>ldapsearchattribute=uid</literal>.
2013 </para>
2014
2015 <para>
2016 If <productname>PostgreSQL</productname> was compiled with
2017 <productname>OpenLDAP</productname> as the LDAP client library, the
2018 <literal>ldapserver</literal> setting may be omitted. In that case, a
2019 list of host names and ports is looked up via
2020 <ulink url="https://datatracker.ietf.org/doc/html/rfc2782">RFC 2782</ulink> DNS SRV records.
2021 The name <literal>_ldap._tcp.DOMAIN</literal> is looked up, where
2022 <literal>DOMAIN</literal> is extracted from <literal>ldapbasedn</literal>.
2023 </para>
2024
2025 <para>
2026 Here is an example for a simple-bind LDAP configuration:
2027 <programlisting>
2028 host ... ldap ldapserver=ldap.example.net ldapprefix="cn=" ldapsuffix=", dc=example, dc=net"
2029 </programlisting>
2030 When a connection to the database server as database
2031 user <literal>someuser</literal> is requested, PostgreSQL will attempt to
2032 bind to the LDAP server using the DN <literal>cn=someuser, dc=example,
2033 dc=net</literal> and the password provided by the client. If that connection
2034 succeeds, the database access is granted.
2035 </para>
2036
2037 <para>
2038 Here is a different simple-bind configuration, which uses the LDAPS scheme
2039 and a custom port number, written as a URL:
2040 <programlisting>
2041 host ... ldap ldapurl="ldaps://ldap.example.net:49151" ldapprefix="cn=" ldapsuffix=", dc=example, dc=net"
2042 </programlisting>
2043 This is slightly more compact than specifying <literal>ldapserver</literal>,
2044 <literal>ldapscheme</literal>, and <literal>ldapport</literal> separately.
2045 </para>
2046
2047 <para>
2048 Here is an example for a search+bind configuration:
2049 <programlisting>
2050 host ... ldap ldapserver=ldap.example.net ldapbasedn="dc=example, dc=net" ldapsearchattribute=uid
2051 </programlisting>
2052 When a connection to the database server as database
2053 user <literal>someuser</literal> is requested, PostgreSQL will attempt to
2054 bind anonymously (since <literal>ldapbinddn</literal> was not specified) to
2055 the LDAP server, perform a search for <literal>(uid=someuser)</literal>
2056 under the specified base DN. If an entry is found, it will then attempt to
2057 bind using that found information and the password supplied by the client.
2058 If that second bind succeeds, the database access is granted.
2059 </para>
2060
2061 <para>
2062 Here is the same search+bind configuration written as a URL:
2063 <programlisting>
2064 host ... ldap ldapurl="ldap://ldap.example.net/dc=example,dc=net?uid?sub"
2065 </programlisting>
2066 Some other software that supports authentication against LDAP uses the
2067 same URL format, so it will be easier to share the configuration.
2068 </para>
2069
2070 <para>
2071 Here is an example for a search+bind configuration that uses
2072 <literal>ldapsearchfilter</literal> instead of
2073 <literal>ldapsearchattribute</literal> to allow authentication by
2074 user ID or email address:
2075 <programlisting>
2076 host ... ldap ldapserver=ldap.example.net ldapbasedn="dc=example, dc=net" ldapsearchfilter="(|(uid=$username)(mail=$username))"
2077 </programlisting>
2078 </para>
2079
2080 <para>
2081 Here is an example for a search+bind configuration that uses DNS SRV
2082 discovery to find the host name(s) and port(s) for the LDAP service for the
2083 domain name <literal>example.net</literal>:
2084 <programlisting>
2085 host ... ldap ldapbasedn="dc=example,dc=net"
2086 </programlisting>
2087 </para>
2088
2089 <tip>
2090 <para>
2091 Since LDAP often uses commas and spaces to separate the different
2092 parts of a DN, it is often necessary to use double-quoted parameter
2093 values when configuring LDAP options, as shown in the examples.
2094 </para>
2095 </tip>
2096
2097 </sect1>
2098
2099 <sect1 id="auth-radius">
2100 <title>RADIUS Authentication</title>
2101
2102 <indexterm zone="auth-radius">
2103 <primary>RADIUS</primary>
2104 </indexterm>
2105
2106 <para>
2107 This authentication method operates similarly to
2108 <literal>password</literal> except that it uses RADIUS
2109 as the password verification method. RADIUS is used only to validate
2110 the user name/password pairs. Therefore the user must already
2111 exist in the database before RADIUS can be used for
2112 authentication.
2113 </para>
2114
2115 <para>
2116 When using RADIUS authentication, an Access Request message will be sent
2117 to the configured RADIUS server. This request will be of type
2118 <literal>Authenticate Only</literal>, and include parameters for
2119 <literal>user name</literal>, <literal>password</literal> (encrypted) and
2120 <literal>NAS Identifier</literal>. The request will be encrypted using
2121 a secret shared with the server. The RADIUS server will respond to
2122 this request with either <literal>Access Accept</literal> or
2123 <literal>Access Reject</literal>. There is no support for RADIUS accounting.
2124 </para>
2125
2126 <para>
2127 Multiple RADIUS servers can be specified, in which case they will
2128 be tried sequentially. If a negative response is received from
2129 a server, the authentication will fail. If no response is received,
2130 the next server in the list will be tried. To specify multiple
2131 servers, separate the server names with commas and surround the list
2132 with double quotes. If multiple servers are specified, the other
2133 RADIUS options can also be given as comma-separated lists, to provide
2134 individual values for each server. They can also be specified as
2135 a single value, in which case that value will apply to all servers.
2136 </para>
2137
2138 <para>
2139 The following configuration options are supported for RADIUS:
2140 <variablelist>
2141 <varlistentry>
2142 <term><literal>radiusservers</literal></term>
2143 <listitem>
2144 <para>
2145 The DNS names or IP addresses of the RADIUS servers to connect to.
2146 This parameter is required.
2147 </para>
2148 </listitem>
2149 </varlistentry>
2150
2151 <varlistentry>
2152 <term><literal>radiussecrets</literal></term>
2153 <listitem>
2154 <para>
2155 The shared secrets used when talking securely to the RADIUS
2156 servers. This must have exactly the same value on the PostgreSQL
2157 and RADIUS servers. It is recommended that this be a string of
2158 at least 16 characters. This parameter is required.
2159 <note>
2160 <para>
2161 The encryption vector used will only be cryptographically
2162 strong if <productname>PostgreSQL</productname> is built with support for
2163 <productname>OpenSSL</productname>. In other cases, the transmission to the
2164 RADIUS server should only be considered obfuscated, not secured, and
2165 external security measures should be applied if necessary.
2166 </para>
2167 </note>
2168 </para>
2169 </listitem>
2170 </varlistentry>
2171
2172 <varlistentry>
2173 <term><literal>radiusports</literal></term>
2174 <listitem>
2175 <para>
2176 The port numbers to connect to on the RADIUS servers. If no port
2177 is specified, the default RADIUS port (<literal>1812</literal>)
2178 will be used.
2179 </para>
2180 </listitem>
2181 </varlistentry>
2182
2183 <varlistentry>
2184 <term><literal>radiusidentifiers</literal></term>
2185 <listitem>
2186 <para>
2187 The strings to be used as <literal>NAS Identifier</literal> in the
2188 RADIUS requests. This parameter can be used, for example, to
2189 identify which database cluster the user is attempting to connect
2190 to, which can be useful for policy matching on
2191 the RADIUS server. If no identifier is specified, the default
2192 <literal>postgresql</literal> will be used.
2193 </para>
2194 </listitem>
2195 </varlistentry>
2196
2197 </variablelist>
2198 </para>
2199
2200 <para>
2201 If it is necessary to have a comma or whitespace in a RADIUS parameter
2202 value, that can be done by putting double quotes around the value, but
2203 it is tedious because two layers of double-quoting are now required.
2204 An example of putting whitespace into RADIUS secret strings is:
2205 <programlisting>
2206 host ... radius radiusservers="server1,server2" radiussecrets="""secret one"",""secret two"""
2207 </programlisting>
2208 </para>
2209 </sect1>
2210
2211 <sect1 id="auth-cert">
2212 <title>Certificate Authentication</title>
2213
2214 <indexterm zone="auth-cert">
2215 <primary>Certificate</primary>
2216 </indexterm>
2217
2218 <para>
2219 This authentication method uses SSL client certificates to perform
2220 authentication. It is therefore only available for SSL connections;
2221 see <xref linkend="ssl-openssl-config"/> for SSL configuration instructions.
2222 When using this authentication method, the server will require that
2223 the client provide a valid, trusted certificate. No password prompt
2224 will be sent to the client. The <literal>cn</literal> (Common Name)
2225 attribute of the certificate
2226 will be compared to the requested database user name, and if they match
2227 the login will be allowed. User name mapping can be used to allow
2228 <literal>cn</literal> to be different from the database user name.
2229 </para>
2230
2231 <para>
2232 The following configuration options are supported for SSL certificate
2233 authentication:
2234 <variablelist>
2235 <varlistentry>
2236 <term><literal>map</literal></term>
2237 <listitem>
2238 <para>
2239 Allows for mapping between system and database user names. See
2240 <xref linkend="auth-username-maps"/> for details.
2241 </para>
2242 </listitem>
2243 </varlistentry>
2244 </variablelist>
2245 </para>
2246
2247 <para>
2248 It is redundant to use the <literal>clientcert</literal> option with
2249 <literal>cert</literal> authentication because <literal>cert</literal>
2250 authentication is effectively <literal>trust</literal> authentication
2251 with <literal>clientcert=verify-full</literal>.
2252 </para>
2253 </sect1>
2254
2255 <sect1 id="auth-pam">
2256 <title>PAM Authentication</title>
2257
2258 <indexterm zone="auth-pam">
2259 <primary>PAM</primary>
2260 </indexterm>
2261
2262 <para>
2263 This authentication method operates similarly to
2264 <literal>password</literal> except that it uses PAM (Pluggable
2265 Authentication Modules) as the authentication mechanism. The
2266 default PAM service name is <literal>postgresql</literal>.
2267 PAM is used only to validate user name/password pairs and optionally the
2268 connected remote host name or IP address. Therefore the user must already
2269 exist in the database before PAM can be used for authentication. For more
2270 information about PAM, please read the
2271 <ulink url="https://www.kernel.org/pub/linux/libs/pam/">
2272 <productname>Linux-PAM</productname> Page</ulink>.
2273 </para>
2274
2275 <para>
2276 The following configuration options are supported for PAM:
2277 <variablelist>
2278 <varlistentry>
2279 <term><literal>pamservice</literal></term>
2280 <listitem>
2281 <para>
2282 PAM service name.
2283 </para>
2284 </listitem>
2285 </varlistentry>
2286 <varlistentry>
2287 <term><literal>pam_use_hostname</literal></term>
2288 <listitem>
2289 <para>
2290 Determines whether the remote IP address or the host name is provided
2291 to PAM modules through the <symbol>PAM_RHOST</symbol> item. By
2292 default, the IP address is used. Set this option to 1 to use the
2293 resolved host name instead. Host name resolution can lead to login
2294 delays. (Most PAM configurations don't use this information, so it is
2295 only necessary to consider this setting if a PAM configuration was
2296 specifically created to make use of it.)
2297 </para>
2298 </listitem>
2299 </varlistentry>
2300 </variablelist>
2301 </para>
2302
2303 <note>
2304 <para>
2305 If PAM is set up to read <filename>/etc/shadow</filename>, authentication
2306 will fail because the PostgreSQL server is started by a non-root
2307 user. However, this is not an issue when PAM is configured to use
2308 LDAP or other authentication methods.
2309 </para>
2310 </note>
2311 </sect1>
2312
2313 <sect1 id="auth-bsd">
2314 <title>BSD Authentication</title>
2315
2316 <indexterm zone="auth-bsd">
2317 <primary>BSD Authentication</primary>
2318 </indexterm>
2319
2320 <para>
2321 This authentication method operates similarly to
2322 <literal>password</literal> except that it uses BSD Authentication
2323 to verify the password. BSD Authentication is used only
2324 to validate user name/password pairs. Therefore the user's role must
2325 already exist in the database before BSD Authentication can be used
2326 for authentication. The BSD Authentication framework is currently
2327 only available on OpenBSD.
2328 </para>
2329
2330 <para>
2331 BSD Authentication in <productname>PostgreSQL</productname> uses
2332 the <literal>auth-postgresql</literal> login type and authenticates with
2333 the <literal>postgresql</literal> login class if that's defined
2334 in <filename>login.conf</filename>. By default that login class does not
2335 exist, and <productname>PostgreSQL</productname> will use the default login class.
2336 </para>
2337
2338 <note>
2339 <para>
2340 To use BSD Authentication, the PostgreSQL user account (that is, the
2341 operating system user running the server) must first be added to
2342 the <literal>auth</literal> group. The <literal>auth</literal> group
2343 exists by default on OpenBSD systems.
2344 </para>
2345 </note>
2346 </sect1>
2347
2348 <sect1 id="auth-oauth">
2349 <title>OAuth Authorization/Authentication</title>
2350
2351 <indexterm zone="auth-oauth">
2352 <primary>OAuth Authorization/Authentication</primary>
2353 </indexterm>
2354
2355 <para>
2356 OAuth 2.0 is an industry-standard framework, defined in
2357 <ulink url="https://datatracker.ietf.org/doc/html/rfc6749">RFC 6749</ulink>,
2358 to enable third-party applications to obtain limited access to a protected
2359 resource.
2360
2361 OAuth client support has to be enabled when <productname>PostgreSQL</productname>
2362 is built, see <xref linkend="installation"/> for more information.
2363 </para>
2364
2365 <para>
2366 This documentation uses the following terminology when discussing the OAuth
2367 ecosystem:
2368
2369 <variablelist>
2370
2371 <varlistentry>
2372 <term>Resource Owner (or End User)</term>
2373 <listitem>
2374 <para>
2375 The user or system who owns protected resources and can grant access to
2376 them. This documentation also uses the term <emphasis>end user</emphasis>
2377 when the resource owner is a person. When you use
2378 <application>psql</application> to connect to the database using OAuth,
2379 you are the resource owner/end user.
2380 </para>
2381 </listitem>
2382 </varlistentry>
2383
2384 <varlistentry>
2385 <term>Client</term>
2386 <listitem>
2387 <para>
2388 The system which accesses the protected resources using access
2389 tokens. Applications using libpq, such as <application>psql</application>,
2390 are the OAuth clients when connecting to a
2391 <productname>PostgreSQL</productname> cluster.
2392 </para>
2393 </listitem>
2394 </varlistentry>
2395
2396 <varlistentry>
2397 <term>Resource Server</term>
2398 <listitem>
2399 <para>
2400 The system hosting the protected resources which are
2401 accessed by the client. The <productname>PostgreSQL</productname>
2402 cluster being connected to is the resource server.
2403 </para>
2404 </listitem>
2405 </varlistentry>
2406
2407 <varlistentry>
2408 <term>Provider</term>
2409 <listitem>
2410 <para>
2411 The organization, product vendor, or other entity which develops and/or
2412 administers the OAuth authorization servers and clients for a given application.
2413 Different providers typically choose different implementation details
2414 for their OAuth systems; a client of one provider is not generally
2415 guaranteed to have access to the servers of another.
2416 </para>
2417 <para>
2418 This use of the term "provider" is not standard, but it seems to be in
2419 wide use colloquially. (It should not be confused with OpenID's similar
2420 term "Identity Provider". While the implementation of OAuth in
2421 <productname>PostgreSQL</productname> is intended to be interoperable
2422 and compatible with OpenID Connect/OIDC, it is not itself an OIDC client
2423 and does not require its use.)
2424 </para>
2425 </listitem>
2426 </varlistentry>
2427
2428 <varlistentry>
2429 <term>Authorization Server</term>
2430 <listitem>
2431 <para>
2432 The system which receives requests from, and issues access tokens to,
2433 the client after the authenticated resource owner has given approval.
2434 <productname>PostgreSQL</productname> does not provide an authorization
2435 server; it is the responsibility of the OAuth provider.
2436 </para>
2437 </listitem>
2438 </varlistentry>
2439
2440 <varlistentry>
2441 <term id="auth-oauth-issuer">Issuer</term>
2442 <listitem>
2443 <para>
2444 An identifier for an authorization server, printed as an
2445 <literal>https://</literal> URL, which provides a trusted "namespace"
2446 for OAuth clients and applications. The issuer identifier allows a
2447 single authorization server to talk to the clients of mutually
2448 untrusting entities, as long as they maintain separate issuers.
2449 </para>
2450 </listitem>
2451 </varlistentry>
2452
2453 </variablelist>
2454
2455 <note>
2456 <para>
2457 For small deployments, there may not be a meaningful distinction between
2458 the "provider", "authorization server", and "issuer". However, for more
2459 complicated setups, there may be a one-to-many (or many-to-many)
2460 relationship: a provider may rent out multiple issuer identifiers to
2461 separate tenants, then provide multiple authorization servers, possibly
2462 with different supported feature sets, to interact with their clients.
2463 </para>
2464 </note>
2465 </para>
2466
2467 <para>
2468 <productname>PostgreSQL</productname> supports bearer tokens, defined in
2469 <ulink url="https://datatracker.ietf.org/doc/html/rfc6750">RFC 6750</ulink>,
2470 which are a type of access token used with OAuth 2.0 where the token is an
2471 opaque string. The format of the access token is implementation specific
2472 and is chosen by each authorization server.
2473 </para>
2474
2475 <para>
2476 The following configuration options are supported for OAuth:
2477 <variablelist>
2478 <varlistentry>
2479 <term><literal>issuer</literal></term>
2480 <listitem>
2481 <para>
2482 An HTTPS URL which is either the exact
2483 <link linkend="auth-oauth-issuer">issuer identifier</link> of the
2484 authorization server, as defined by its discovery document, or a
2485 well-known URI that points directly to that discovery document. This
2486 parameter is required.
2487 </para>
2488 <para>
2489 When an OAuth client connects to the server, a URL for the discovery
2490 document will be constructed using the issuer identifier. By default,
2491 this URL uses the conventions of OpenID Connect Discovery: the path
2492 <literal>/.well-known/openid-configuration</literal> will be appended
2493 to the end of the issuer identifier. Alternatively, if the
2494 <literal>issuer</literal> contains a <literal>/.well-known/</literal>
2495 path segment, that URL will be provided to the client as-is.
2496 </para>
2497 <warning>
2498 <para>
2499 The OAuth client in libpq requires the server's issuer setting to
2500 exactly match the issuer identifier which is provided in the discovery
2501 document, which must in turn match the client's
2502 <xref linkend="libpq-connect-oauth-issuer"/> setting. No variations in
2503 case or formatting are permitted.
2504 </para>
2505 </warning>
2506 </listitem>
2507 </varlistentry>
2508
2509 <varlistentry>
2510 <term><literal>scope</literal></term>
2511 <listitem>
2512 <para>
2513 A space-separated list of the OAuth scopes needed for the server to
2514 both authorize the client and authenticate the user. Appropriate values
2515 are determined by the authorization server and the OAuth validation
2516 module used (see <xref linkend="oauth-validators" /> for more
2517 information on validators). This parameter is required.
2518 </para>
2519 </listitem>
2520 </varlistentry>
2521
2522 <varlistentry>
2523 <term><literal>validator</literal></term>
2524 <listitem>
2525 <para>
2526 The library to use for validating bearer tokens. If given, the name must
2527 exactly match one of the libraries listed in
2528 <xref linkend="guc-oauth-validator-libraries" />. This parameter is
2529 optional unless <literal>oauth_validator_libraries</literal> contains
2530 more than one library, in which case it is required.
2531 </para>
2532 </listitem>
2533 </varlistentry>
2534
2535 <varlistentry>
2536 <term><literal>map</literal></term>
2537 <listitem>
2538 <para>
2539 Allows for mapping between OAuth identity provider and database user
2540 names. See <xref linkend="auth-username-maps"/> for details. If a
2541 map is not specified, the user name associated with the token (as
2542 determined by the OAuth validator) must exactly match the role name
2543 being requested. This parameter is optional.
2544 </para>
2545 </listitem>
2546 </varlistentry>
2547
2548 <varlistentry>
2549 <term id="auth-oauth-delegate-ident-mapping" xreflabel="delegate_ident_mapping">
2550 <literal>delegate_ident_mapping</literal>
2551 </term>
2552 <listitem>
2553 <para>
2554 An advanced option which is not intended for common use.
2555 </para>
2556 <para>
2557 When set to <literal>1</literal>, standard user mapping with
2558 <filename>pg_ident.conf</filename> is skipped, and the OAuth validator
2559 takes full responsibility for mapping end user identities to database
2560 roles. If the validator authorizes the token, the server trusts that
2561 the user is allowed to connect under the requested role, and the
2562 connection is allowed to proceed regardless of the authentication
2563 status of the user.
2564 </para>
2565 <para>
2566 This parameter is incompatible with <literal>map</literal>.
2567 </para>
2568 <warning>
2569 <para>
2570 <literal>delegate_ident_mapping</literal> provides additional
2571 flexibility in the design of the authentication system, but it also
2572 requires careful implementation of the OAuth validator, which must
2573 determine whether the provided token carries sufficient end-user
2574 privileges in addition to the <link linkend="oauth-validators">standard
2575 checks</link> required of all validators. Use with caution.
2576 </para>
2577 </warning>
2578 </listitem>
2579 </varlistentry>
2580 </variablelist>
2581 </para>
2582 </sect1>
2583
2584 <sect1 id="client-authentication-problems">
2585 <title>Authentication Problems</title>
2586
2587 <para>
2588 Authentication failures and related problems generally
2589 manifest themselves through error messages like the following:
2590 </para>
2591
2592 <para>
2593 <programlisting>
2594 FATAL: no pg_hba.conf entry for host "123.123.123.123", user "andym", database "testdb"
2595 </programlisting>
2596 This is what you are most likely to get if you succeed in contacting
2597 the server, but it does not want to talk to you. As the message
2598 suggests, the server refused the connection request because it found
2599 no matching entry in its <filename>pg_hba.conf</filename>
2600 configuration file.
2601 </para>
2602
2603 <para>
2604 <programlisting>
2605 FATAL: password authentication failed for user "andym"
2606 </programlisting>
2607 Messages like this indicate that you contacted the server, and it is
2608 willing to talk to you, but not until you pass the authorization
2609 method specified in the <filename>pg_hba.conf</filename> file. Check
2610 the password you are providing, or check your Kerberos or ident
2611 software if the complaint mentions one of those authentication
2612 types.
2613 </para>
2614
2615 <para>
2616 <programlisting>
2617 FATAL: user "andym" does not exist
2618 </programlisting>
2619 The indicated database user name was not found.
2620 </para>
2621
2622 <para>
2623 <programlisting>
2624 FATAL: database "testdb" does not exist
2625 </programlisting>
2626 The database you are trying to connect to does not exist. Note that
2627 if you do not specify a database name, it defaults to the database
2628 user name.
2629 </para>
2630
2631 <tip>
2632 <para>
2633 The server log might contain more information about an
2634 authentication failure than is reported to the client. If you are
2635 confused about the reason for a failure, check the server log.
2636 </para>
2637 </tip>
2638 </sect1>
2639
2640 </chapter>