runtime.sgml 85.2 KB
Newer Older
1
<!--
2
$Header: /cvsroot/pgsql/doc/src/sgml/runtime.sgml,v 1.78 2001/09/12 14:06:37 petere Exp $
3 4
-->

5 6 7 8 9 10 11 12 13 14 15
<Chapter Id="runtime">
 <Title>Server Runtime Environment</Title>

 <Para>
  This chapter discusses how to set up and run the database server
  and the interactions with the operating system.
 </para>

 <sect1 id="postgres-user">
  <title>The Postgres user account</title>

16 17 18 19
  <indexterm>
   <primary>postgres user</primary>
  </indexterm>

20 21 22 23 24 25 26 27 28 29 30
  <para>
   As with any other server daemon that is connected to the world at
   large, it is advisable to run Postgres under a separate user
   account. This user account should only own the data itself that is
   being managed by the server, and should not be shared with other
   daemons. (Thus, using the user <quote>nobody</quote> is a bad
   idea.) It is not advisable to install the executables as owned by
   this user account because that runs the risk of user-defined
   functions gone astray or any other exploits compromising the
   executable programs.
  </para>
31

32 33 34
  <para>
   To add a user account to your system, look for a command
   <command>useradd</command> or <command>adduser</command>. The user
35
   name <systemitem>postgres</systemitem> is often used but by no means
36 37 38 39 40 41 42
   required.
  </para>
 </sect1>

 <sect1 id="creating-cluster">
  <title>Creating a database cluster</title>

43 44 45 46 47 48 49 50 51
  <indexterm>
   <primary>database cluster</primary>
  </indexterm>

  <indexterm>
   <primary>data area</primary>
   <see>database cluster</see>
  </indexterm>

52 53 54 55 56 57 58 59 60 61
  <para>
   Before you can do anything, you must initialize a database storage
   area on disk. We call this a <firstterm>database
   cluster</firstterm>. (<acronym>SQL</acronym> speaks of a catalog
   cluster instead.) A database cluster is a collection of databases
   that will be accessible through a single instance of a running
   database server. After initialization, a database cluster will
   contain one database named <literal>template1</literal>. As the
   name suggests, this will be used as a template for any subsequently
   created database; it should not be used for actual work.
62 63
  </para>

64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82
  <para>
   In file system terms, a database cluster will be a single directory
   under which all data will be stored. We call this the
   <firstterm>data directory</firstterm> or <firstterm>data
   area</firstterm>. It is completely up to you where you choose to
   store your data, there is no default, although locations such as
   <filename>/usr/local/pgsql/data</filename> or
   <filename>/var/lib/pgsql/data</filename> are popular. To initialize
   a database cluster, use the command <command>initdb</command>,
   which is installed with <productname>PostgreSQL</productname>. The
   desired file system location of your database system is indicated
   by the <option>-D</option> option, for example
<screen>
&gt; <userinput>initdb -D /usr/local/pgsql/data</userinput>
</screen>
   Note that you must execute this command while being logged in to
   the Postgres user account, which is described in the previous
   section.
  </para>
83

84
  <tip>
85
   <para>
86 87 88
    <indexterm>
     <primary><envar>PGDATA</envar></primary>
    </indexterm>
89 90
    As an alternative to the <option>-D</option> option, you can set
    the environment variable <envar>PGDATA</envar>.
91
   </para>
92 93 94 95 96 97
  </tip>

  <para>
   <command>initdb</command> will attempt to create the directory you
   specify if it does not already exist. It is likely that it won't
   have the permission to do so (if you followed our advice and
98 99 100
   created an unprivileged account). In that case you should create the
   directory yourself (as root) and transfer ownership of it to the
   Postgres user account. Here is how this might work:
101 102 103 104 105 106 107
<screen>
root# <userinput>mkdir /usr/local/pgsql/data</userinput>
root# <userinput>chown postgres /usr/local/pgsql/data</userinput>
root# <userinput>su postgres</userinput>
postgres&gt; <userinput>initdb -D /usr/local/pgsql/data</userinput>
</screen>
  </para>
108

109 110 111 112 113 114 115 116 117 118 119
  <para>
   <command>initdb</command> will refuse to run if the data directory
   looks like it belongs to an already initialized installation.
  </para>

  <para>
   Because the data directory contains all the data stored in the
   database it is essential that it be well secured from unauthorized
   access. <command>initdb</command> therefore revokes access
   permissions from everyone but the Postgres user account.
  </para>
120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138

  <para>
   One surprise you might encounter while running <command>initdb</command> is
   a notice similar to this one:
<screen>
NOTICE:  Initializing database with en_US collation order.
        This locale setting will prevent use of index optimization for
        LIKE and regexp searches.  If you are concerned about speed of
        such queries, you may wish to set LC_COLLATE to "C" and
        re-initdb.  For more information see the Administrator's Guide.
</screen>
   This notice is intended to warn you that the currently selected locale
   will cause indexes to be sorted in an order that prevents them from
   being used for LIKE and regular-expression searches.  If you need
   good performance of such searches, you should set your current locale
   to "C" and re-run <command>initdb</command>.  On most systems, setting the
   current locale is done by changing the value of the environment variable
   <literal>LC_ALL</literal> or <literal>LANG</literal>.  The sort order used
   within a particular database cluster is set by <command>initdb</command>
139
   and cannot be changed later, short of dumping all data, rerunning <command>initdb</command>,
140 141
   reload data.  So it's important to make this choice correctly now.
  </para>
142 143 144 145 146 147
 </sect1>

 <sect1 id="postmaster-start">
  <title>Starting the database server</title>

  <para>
148 149 150
   <indexterm>
    <primary>postmaster</primary>
   </indexterm>
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
   Before anyone can access the database you must start the database
   server. The database server is called
   <firstterm>postmaster</firstterm>.
   The postmaster must know where to find the data it is supposed
   to work on. This is done with the <option>-D</option> option. Thus,
   the simplest way to start the server is, for example,
<screen>
&gt; <userinput>postmaster -D /usr/local/pgsql/data</userinput>
</screen>
   which will leave the server running in the foreground. This must
   again be done while logged in to the Postgres user account. Without
   a <option>-D</option>, the server will try to use the data
   directory in the environment variable <envar>PGDATA</envar>; if
   neither of these works it will fail.
  </para>

  <para>
   To start the <application>postmaster</application> in the
   background, use the usual shell syntax:
<screen>
171
&gt; <userinput>postmaster -D /usr/local/pgsql/data &gt; logfile 2&gt;&amp;1 &amp;</userinput>
172 173 174 175 176 177 178
</screen>
   It is an extremely good idea to keep the server output around
   somewhere, as indicated here. It will help both for auditing
   purposes and to diagnose problems.
  </para>

  <para>
179 180 181
   <indexterm>
    <primary>TCP/IP</primary>
   </indexterm>
182 183 184 185 186 187 188
   The postmaster also takes a number of other command line options.
   For more information see the reference page and below under runtime
   configuration. In particular, in order for the postmaster to accept
   TCP/IP connections (rather than just Unix domain socket ones), you
   must also specify the <option>-i</option> option.
  </para>

189
  <para>
190 191 192
   <indexterm>
    <primary>pg_ctl</primary>
   </indexterm>
193
   This shell syntax can get tedious quickly.  Therefore the shell
194 195 196 197 198 199 200 201 202 203 204 205
   script wrapper <application>pg_ctl</application> is provided that
   encapsulates some of the tasks.  E.g.,
<programlisting>
pg_ctl start -l logfile
</programlisting>
   will start the server in the background and put the output into the
   named log file.  The <option>-D</option> option has the same
   meaning as when invoking postmaster directly.
   <application>pg_ctl</application> also implements a symmetric
   <quote>stop</quote> operation.
  </para>

206 207 208 209 210 211 212 213 214 215 216 217 218 219
  <para>
   Normally, you will want to start the database server when the
   computer boots up. This is not required; the
   <productname>PostgreSQL</productname> server can be run
   successfully from non-privileged accounts without root
   intervention.
  </para>

  <para>
   Different systems have different conventions for starting up
   daemons at boot time, so you are advised to familiarize yourself
   with them. Many systems have a file
   <filename>/etc/rc.local</filename> or
   <filename>/etc/rc.d/rc.local</filename> which is almost certainly
220
   no bad place to put such a command. Whatever you do, the server
221 222 223 224 225
   must be run by the <productname>Postgres</productname> user account
   <emphasis>and not by root</emphasis> or any other user. Therefore
   you probably always want to form your command lines along the lines
   of <literal>su -c '...' postgres</literal>, for example:
<programlisting>
226
su -c 'pg_ctl start -D /usr/local/pgsql/data -l serverlog' postgres
227 228 229 230
</programlisting>
  </para>

  <para>
231 232 233
   Here are a few more operating system specific suggestions.  (Always
   replace the proper installation directory and the user name you
   chose.)
234 235 236 237

   <itemizedlist>
    <listitem>
     <para>
238 239 240 241 242 243 244 245 246 247
      For <productname>FreeBSD</productname>, take a look at the file
      <filename>contrib/start-scripts/freebsd</filename> in the
      <productname>PostgreSQL</productname> source distribution.
     </para>
    </listitem>

    <listitem>
     <para>
      On <productname>OpenBSD</productname>, add the following lines
      to the file <filename>/etc/rc.local</filename>:
248
<programlisting>
249 250 251 252
if [ -x /usr/local/pgsql/bin/pg_ctl -a -x /usr/local/pgsql/bin/postmaster ]; then
    su - -c '/usr/local/pgsql/bin/pg_ctl start -l /var/postgresql/log -s' postgres
    echo -n ' postgresql'
fi
253 254 255 256 257 258
</programlisting>
     </para>
    </listitem>

    <listitem>
     <para>
259
      On <productname>Linux</productname> systems either add
260
<programlisting>
261
/usr/local/pgsql/bin/pg_ctl start -l logfile -D /usr/local/pgsql/data
262
</programlisting>
263 264 265 266
      to <filename>/etc/rc.d/rc.local</filename> or look into the file
      <filename>contrib/start-scripts/linux</filename> in the
      <productname>PostgreSQL</productname> source distribution to
      integrate the start and shutdown into the run level system.
267 268 269 270 271
     </para>
    </listitem>

    <listitem>
     <para>
272 273 274 275 276 277 278 279 280 281
      On <productname>NetBSD</productname>, either use the
      <productname>FreeBSD</productname> or
      <productname>Linux</productname> start scripts, depending on
      preference, as an example and place the file at
      <filename>/usr/local/etc/rc.d/postgresql</filename>.
     </para>
    </listitem>

    <listitem>
     <para>
282 283 284
      On <productname>Solaris</productname>, create a file called
      <filename>/etc/init.d/postgresql</filename> to contain the following 
      single line:
285 286 287
<programlisting>
su - postgres -c "/usr/local/pgsql/bin/pg_ctl start -l logfile -D /usr/local/pgsql/data"
</programlisting>
288
      Then, create a symbolic link to it in <filename>/etc/rc3.d</> as
289
      <literal>S99postgresql</>.
290 291 292 293 294
     </para>
    </listitem>
   </itemizedlist>

  </para>
295

296
   <para>
297
    While the <application>postmaster</application> is running, its
298
    <acronym>PID</acronym> is in the file <filename>postmaster.pid</filename> in the data
299 300
    directory. This is used as an interlock against multiple postmasters
    running in the same data directory, and can also be used for
301
    shutting down the postmaster.
302 303
   </para>

304
   <sect2 id="postmaster-start-failures">
305
    <title>Server Start-up Failures</title>
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337

    <para>
     There are several common reasons for the postmaster to fail to
     start up. Check the postmaster's log file, or start it by hand
     (without redirecting standard output or standard error) to see
     what complaint messages appear. Some of the possible error
     messages are reasonably self-explanatory, but here are some that
     are not.
    </para>

    <para>
<screen>
FATAL: StreamServerPort: bind() failed: Address already in use
        Is another postmaster already running on that port?
</screen>
     This usually means just what it suggests: you accidentally
     started a second postmaster on the same port where one is already
     running. However, if the kernel error message is not
     <computeroutput>Address already in use</computeroutput> or some
     variant of that wording, there may be a different problem. For
     example, trying to start a postmaster on a reserved port number
     may draw something like
<screen>
&gt; <userinput>postmaster -i -p 666</userinput>
FATAL: StreamServerPort: bind() failed: Permission denied
        Is another postmaster already running on that port?
</screen>
    </para>

    <para>
     A message like
<screen>
338
IpcMemoryCreate: shmget(key=5440001, size=83918612, 01600) failed: Invalid argument
339 340 341 342 343
FATAL 1:  ShmemCreate: cannot create region
</screen>
     probably means that your kernel's limit on the size of shared
     memory areas is smaller than the buffer area that Postgres is
     trying to create (83918612 bytes in this example). Or it could
344
     mean that you don't have System-V-style shared memory support
345 346 347 348 349 350 351 352 353 354 355 356
     configured into your kernel at all. As a temporary workaround,
     you can try starting the postmaster with a smaller-than-normal
     number of buffers (<option>-B</option> switch). You will
     eventually want to reconfigure your kernel to increase the
     allowed shared memory size, however. You may see this message
     when trying to start multiple postmasters on the same machine, if
     their total space requests exceed the kernel limit.
    </para>

    <para>
     An error like
<screen>
357
IpcSemaphoreCreate: semget(key=5440026, num=16, 01600) failed: No space left on device
358 359
</screen>
     does <emphasis>not</emphasis> mean that you've run out of disk
360 361
     space; it means that your kernel's limit on the number of System
     V semaphores is smaller than the number
362 363 364 365 366 367
     <productname>Postgres</productname> wants to create. As above,
     you may be able to work around the problem by starting the
     postmaster with a reduced number of backend processes
     (<option>-N</option> switch), but you'll eventually want to
     increase the kernel limit.
    </para>
368

Peter Eisentraut's avatar
Peter Eisentraut committed
369 370 371 372 373 374 375
    <para>
     If you get an <quote>illegal system call</> error, then it is likely that
     shared memory or semaphores are not supported at all in your kernel. In
     that case your only option is to re-configure the kernel to turn on these
     features.
    </para>

376
    <para>
377
     Details about configuring <systemitem class="osname">System V</> <acronym>IPC</> facilities are given in
378 379
     <xref linkend="sysvipc">.
    </para>
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394
   </sect2>

   <sect2 id="client-connection-problems">
    <title>Client Connection Problems</title>

    <para>
     Although the possible error conditions on the client side are
     both virtually infinite and application dependent, a few of them
     might be directly related to how the server was started up.
     Conditions other than those shown below should be documented with
     the respective client application.
    </para>

    <para>
<screen>
395 396 397
psql: could not connect to server: Connection refused
        Is the server running on host server.joe.com and accepting
        TCP/IP connections on port 5432?
398 399 400 401 402 403 404 405 406 407 408 409
</screen>
     This is the generic <quote>I couldn't find a server to talk
     to</quote> failure. It looks like the above when TCP/IP
     communication is attempted. A common mistake is to forget the
     <option>-i</option> to the postmaster to allow TCP/IP
     connections.
    </para>

    <para>
     Alternatively, you'll get this when attempting
     Unix-socket communication to a local postmaster:
<screen>
410 411 412
psql: could not connect to server: Connection refused
        Is the server running locally and accepting
        connections on Unix domain socket "/tmp/.s.PGSQL.5432"?
413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432
</screen>
    </para>

    <para>
     The last line is useful in verifying that the client is trying to
     connect where it is supposed to. If there is in fact no
     postmaster running there, the kernel error message will typically
     be either <computeroutput>Connection refused</computeroutput> or
     <computeroutput>No such file or directory</computeroutput>, as
     illustrated. (It is particularly important to realize that
     <computeroutput>Connection refused</computeroutput> in this
     context does <emphasis>not</emphasis> mean that the postmaster
     got your connection request and rejected it -- that case will
     produce a different message, as shown in <xref
     linkend="client-authentication-problems">.) Other error messages
     such as <computeroutput>Connection timed out</computeroutput> may
     indicate more fundamental problems, like lack of network
     connectivity.
    </para>
   </sect2>
433 434
  </sect1>

435
  <sect1 id="runtime-config">
436
   <Title>Run-time configuration</Title>
437

438 439 440 441 442
   <indexterm>
    <primary>configuration</primary>
    <secondary>server</secondary>
   </indexterm>

443 444 445 446 447
   <para>
    There are a lot of configuration parameters that affect the
    behavior of the database system in some way or other. Here we
    describe how to set them and the following subsections will
    discuss each of them.
448 449 450
   </para>

   <para>
451
    All parameter names are case-insensitive. Every parameter takes a
452
    value of one of the four types Boolean, integer, floating point,
453 454 455 456 457 458
    string as described below. Boolean values are
    <literal>ON</literal>, <literal>OFF</literal>,
    <literal>TRUE</literal>, <literal>FALSE</literal>,
    <literal>YES</literal>, <literal>NO</literal>,
    <literal>1</literal>, <literal>0</literal> (case-insensitive) or
    any non-ambiguous prefix of these.
459 460
   </para>

461 462 463
   <para>
    One way to set these options is to create a file
    <filename>postgresql.conf</filename> in the data directory (e.g.,
464 465
    <filename>/usr/local/pgsql/data</filename>). An example of what
    this file could look like is:
466 467 468 469 470 471 472 473 474
<programlisting>
# This is a comment
log_connections = yes
syslog = 2
</programlisting>
    As you see, options are one per line. The equal sign between name
    and value is optional. White space is insignificant, blank lines
    are ignored. Hash marks (<quote>#</quote>) introduce comments
    anywhere.
475 476 477
   </para>

   <para>
478 479 480
    <indexterm>
     <primary>SIGHUP</primary>
    </indexterm>
481
    The configuration file is reread whenever the postmaster receives
482
    a <systemitem>SIGHUP</> signal. This signal is also propagated to all running
483 484 485
    backend processes, so that running sessions get the new default.
    Alternatively, you can send the signal to only one backend process
    directly.
486 487 488
   </para>

   <para>
489 490 491
    A second way to set these configuration parameters is to give them
    as a command line option to the postmaster, such as
<programlisting>
492
postmaster -c log_connections=yes -c syslog=2
493 494
</programlisting>
    which would have the same effect as the previous example.
495 496
    Command-line options override any conflicting settings in
    <filename>postgresql.conf</filename>.
497 498
   </para>

499 500 501 502 503 504
   <para>
    Occasionally it is also useful to give a command line option to
    one particular backend session only. The environment variable
    <envar>PGOPTIONS</envar> can be used for this purpose on the
    client side:
<programlisting>
505
env PGOPTIONS='-c geqo=off' psql
506 507 508 509 510 511
</programlisting>
    (This works for any client application, not just
    <application>psql</application>.) Note that this won't work for
    options that are necessarily fixed once the server is started,
    such as the port number.
   </para>
512

513 514 515 516 517 518 519 520
   <para>
    Finally, some options can be changed in individual SQL sessions
    with the <command>SET</command> command, for example
<screen>
=&gt; <userinput>SET ENABLE_SEQSCAN TO OFF;</userinput>
</screen>
    See the SQL command language reference for details on the syntax.
   </para>
521

522 523
   <sect2 id="runtime-config-optimizer">
    <title>Planner and Optimizer Tuning</title>
524

525 526 527
   <para>
    <variablelist>
     <varlistentry>
528
      <term><varname>CPU_INDEX_TUPLE_COST</varname> (<type>floating point</type>)</term>
529 530 531 532 533 534
      <listitem>
       <para>
        Sets the query optimizer's estimate of the cost of processing
	each index tuple during an index scan. This is measured as a
	fraction of the cost of a sequential page fetch.
       </para>
535 536
      </listitem>
     </varlistentry>
537
    
538
     <varlistentry>
539
      <term><varname>CPU_OPERATOR_COST</varname> (<type>floating point</type>)</term>
540 541
      <listitem>
       <para>
542 543 544
        Sets the optimizer's estimate of the cost of processing each
	operator in a WHERE clause. This is measured as a fraction of
	the cost of a sequential page fetch.
545
       </para>
546 547 548 549
      </listitem>
     </varlistentry>
    
     <varlistentry>
550
      <term><varname>CPU_TUPLE_COST</varname> (<type>floating point</type>)</term>
551 552 553 554 555 556 557 558 559 560
      <listitem>
       <para>
        Sets the query optimizer's estimate of the cost of processing
	each tuple during a query. This is measured as a fraction of
	the cost of a sequential page fetch.
       </para>
      </listitem>
     </varlistentry>
    
     <varlistentry>
561
      <term><varname>EFFECTIVE_CACHE_SIZE</varname> (<type>floating point</type>)</term>
562 563 564 565 566 567 568 569 570 571
      <listitem>
       <para>
        Sets the optimizer's assumption about the effective size of
	the disk cache (that is, the portion of the kernel's disk
	cache that will be used for
	<productname>Postgres</productname> data files). This is
	measured in disk pages, which are normally 8kB apiece.
       </para>
      </listitem>
     </varlistentry>
572

573
     <varlistentry>
574
      <term><varname>ENABLE_HASHJOIN</varname> (<type>boolean</type>)</term>
575 576 577 578 579 580
      <listitem>
       <para>
        Enables or disables the query planner's use of hash-join plan
        types. The default is on. This is mostly useful to debug the
        query planner.
       </para>
581 582 583 584
      </listitem>
     </varlistentry>

     <varlistentry>
585 586 587 588
      <indexterm>
       <primary>index scan</primary>
      </indexterm>

589
      <term><varname>ENABLE_INDEXSCAN</varname> (<type>boolean</type>)</term>
590 591
      <listitem>
       <para>
592 593 594
        Enables or disables the query planner's use of index scan plan
        types. The default is on. This is mostly useful to debug the
        query planner.
595 596 597 598 599
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
600
      <term><varname>ENABLE_MERGEJOIN</varname> (<type>boolean</type>)</term>
601 602
      <listitem>
       <para>
603 604 605
        Enables or disables the query planner's use of merge-join plan
        types. The default is on. This is mostly useful to debug the
        query planner.
606 607 608 609 610
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
611
      <term><varname>ENABLE_NESTLOOP</varname> (<type>boolean</type>)</term>
612 613
      <listitem>
       <para>
614 615 616 617 618 619
        Enables or disables the query planner's use of nested-loop
	join plans. It's not possible to suppress nested-loop joins
	entirely, but turning this variable off discourages the
	planner from using one if there is any other method available.
	The default is on. This is mostly useful to debug the query
	planner.
620 621 622 623
       </para>
      </listitem>
     </varlistentry>

624
     <varlistentry>
625 626 627 628
      <indexterm>
       <primary>sequential scan</primary>
      </indexterm>

629
      <term><varname>ENABLE_SEQSCAN</varname> (<type>boolean</type>)</term>
630 631
      <listitem>
       <para>
632 633 634 635 636 637
        Enables or disables the query planner's use of sequential scan
	plan types. It's not possible to suppress sequential scans
	entirely, but turning this variable off discourages the
	planner from using one if there is any other method available.
	The default is on. This is mostly useful to debug the query
	planner.
638 639 640 641 642
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
643
      <term><varname>ENABLE_SORT</varname> (<type>boolean</type>)</term>
644 645
      <listitem>
       <para>
646 647 648 649 650
        Enables or disables the query planner's use of explicit sort
	steps. It's not possible to suppress explicit sorts entirely,
	but turning this variable off discourages the planner from
	using one if there is any other method available. The default
	is on. This is mostly useful to debug the query planner.
651 652 653 654 655
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
656
      <term><varname>ENABLE_TIDSCAN</varname> (<type>boolean</type>)</term>
657 658
      <listitem>
       <para>
659
        Enables or disables the query planner's use of <acronym>TID</> scan plan
660 661
        types. The default is on. This is mostly useful to debug the
        query planner.
662 663 664 665
       </para>
      </listitem>
     </varlistentry>

666
     <varlistentry>
667 668 669 670 671 672 673 674
      <indexterm>
       <primary>genetic query optimization</primary>
      </indexterm>
      <indexterm>
       <primary>GEQO</primary>
       <see>genetic query optimization</see>
      </indexterm>

675
      <term><varname>GEQO</varname> (<type>boolean</type>)</term>
676 677
      <listitem>
       <para>
678 679 680
        Enables or disables genetic query optimization, which is an
        algorithm that attempts to do query planning without
        exhaustive search. This is on by default. See also the various
681
        other <varname>GEQO_</varname> settings.
682 683 684 685 686
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
687 688 689 690 691
      <term><varname>GEQO_EFFORT</varname> (<type>integer</type>)</term>
      <term><varname>GEQO_GENERATIONS</varname> (<type>integer</type>)</term>
      <term><varname>GEQO_POOL_SIZE</varname> (<type>integer</type>)</term>
      <term><varname>GEQO_RANDOM_SEED</varname> (<type>integer</type>)</term>
      <term><varname>GEQO_SELECTION_BIAS</varname> (<type>floating point</type>)</term>
692 693
      <listitem>
       <para>
694 695 696 697
        Various tuning parameters for the genetic query optimization
        algorithm: The pool size is the number of individuals in one
        population. Valid values are between 128 and 1024. If it is
        set to 0 (the default) a pool size of 2^(QS+1), where QS
698
        is the number of FROM items in the query, is taken. The effort
699 700 701 702
        is used to calculate a default for generations. Valid values
        are between 1 and 80, 40 being the default. Generations
        specifies the number of iterations in the algorithm. The
        number must be a positive integer. If 0 is specified then
703
        <literal>Effort * Log2(PoolSize)</literal> is used. The run time of the algorithm
704 705 706 707
        is roughly proportional to the sum of pool size and
        generations. The selection bias is the selective pressure
        within the population. Values can be from 1.50 to 2.00; the
        latter is the default. The random seed can be set to get
708
        reproducible results from the algorithm. If it is set to -1
709
        then the algorithm behaves non-deterministically.
710 711 712 713 714
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
715
      <term><varname>GEQO_THRESHOLD</varname> (<type>integer</type>)</term>
716 717
      <listitem>
       <para>
718 719 720 721
        Use genetic query optimization to plan queries with at least
        this many FROM items involved.  (Note that a JOIN construct
	counts as only one FROM item.) The default is 11. For simpler
	queries it is usually best to use the
722
        deterministic, exhaustive planner.
723 724 725 726 727
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
728
      <term><varname>KSQO</varname> (<type>boolean</type>)</term>
729 730
      <listitem>
       <para>
731
        The <firstterm>Key Set Query Optimizer</firstterm>
732
        (<acronym>KSQO</acronym>) causes the query planner to convert
733 734 735 736 737 738
        queries whose WHERE clause contains many OR'ed AND clauses
        (such as <literal>WHERE (a=1 AND b=2) OR (a=2 AND b=3)
        ...</literal>) into a UNION query. This method can be faster
        than the default implementation, but it doesn't necessarily
        give exactly the same results, since UNION implicitly adds a
        SELECT DISTINCT clause to eliminate identical output rows.
739
        <acronym>KSQO</acronym> is commonly used when working with products like
740 741 742 743 744
        <productname>Microsoft Access</productname>, which tend to
        generate queries of this form.
       </para>

       <para>
745
        The <acronym>KSQO</acronym> algorithm used to be absolutely essential for queries
746 747 748 749
        with many OR'ed AND clauses, but in
        <productname>Postgres</productname> 7.0 and later the standard
        planner handles these queries fairly successfully. Hence the
        default is OFF.
750 751 752 753 754
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
755
      <term><varname>RANDOM_PAGE_COST</varname> (<type>floating point</type>)</term>
756 757
      <listitem>
       <para>
758 759 760
        Sets the query optimizer's estimate of the cost of a
	nonsequentially fetched disk page. This is measured as a
	multiple of the cost of a sequential page fetch.
761 762 763
       </para>
      </listitem>
     </varlistentry>
764 765 766 767 768 769 770 771 772 773 774 775 776
    </variablelist>
   </para>

   <note>
    <para>
     Unfortunately, there is no well-defined method of determining
     ideal values for the family of <quote>COST</quote> variables that
     were just described. You are encouraged to experiment and share
     your findings.
    </para>
   </note>

   </sect2>
777

778 779 780 781 782
   <sect2 id="logging">
    <title>Logging and Debugging</title>

   <para>
    <variablelist>
783
     <varlistentry>
784
      <term><varname>DEBUG_ASSERTIONS</varname> (<type>boolean</type>)</term>
785 786 787 788 789
      <listitem>
       <para>
        Turns on various assertion checks. This is a debugging aid. If
        you are experiencing strange problems or crashes you might
        want to turn this on, as it might expose programming mistakes.
790 791 792 793 794
        To use this option, the macro <literal>USE_ASSERT_CHECKING</literal>
        must be defined when Postgres is built (see the configure option
	<literal>--enable-cassert</literal>).  Note that
	<literal>DEBUG_ASSERTIONS</literal> defaults to ON if Postgres
	has been built this way.
795 796 797 798
       </para>
      </listitem>
     </varlistentry>

799
     <varlistentry>
800
      <term><varname>DEBUG_LEVEL</varname> (<type>integer</type>)</term>
801 802
      <listitem>
       <para>
803 804 805 806 807
        The higher this value is set, the more
        <quote>debugging</quote> output of various sorts is generated
        in the server log during operation. This option is 0 by
        default, which means no debugging output. Values up to about 4
        currently make sense.
808 809 810
       </para>
      </listitem>
     </varlistentry>
811

812
     <varlistentry>
813 814 815 816 817
      <term><varname>DEBUG_PRINT_QUERY</varname> (<type>boolean</type>)</term>
      <term><varname>DEBUG_PRINT_PARSE</varname> (<type>boolean</type>)</term>
      <term><varname>DEBUG_PRINT_REWRITTEN</varname> (<type>boolean</type>)</term>
      <term><varname>DEBUG_PRINT_PLAN</varname> (<type>boolean</type>)</term>
      <term><varname>DEBUG_PRETTY_PRINT</varname> (<type>boolean</type>)</term>
818 819
      <listitem>
       <para>
820 821 822 823 824 825 826
        These flags enable various debugging output to be sent to the
	server log.  For each executed query, prints either the query text,
	the resulting parse tree, the query rewriter output, or the execution
	plan. <option>DEBUG_PRETTY_PRINT</option> indents these displays
	to produce a more readable but much longer output format.
	Setting <option>DEBUG_LEVEL</option> above zero implicitly turns
	on some of these flags.
827 828 829 830 831
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
832
      <term><varname>HOSTNAME_LOOKUP</varname> (<type>boolean</type>)</term>
833 834
      <listitem>
       <para>
835 836 837 838 839
        By default, connection logs only show the IP address of the
        connecting host. If you want it to show the host name you can
        turn this on, but depending on your host name resolution setup
        it might impose a non-negligible performance penalty. This
        option can only be set at server start.
840 841 842 843 844
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
845
      <term><varname>LOG_CONNECTIONS</varname> (<type>boolean</type>)</term>
846 847
      <listitem>
       <para>
848 849 850 851
        Prints a line informing about each successful connection to
        the server log. This is off by default, although it is
        probably very useful. This option can only be set at server
        start.
852 853 854 855 856
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
857
      <term><varname>LOG_PID</varname> (<type>boolean</type>)</term>
858 859
      <listitem>
       <para>
860 861 862
        Prefixes each server log message with the process id of the
        backend process. This is useful to sort out which messages
        pertain to which connection. The default is off.
863 864 865 866 867
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
868
      <term><varname>LOG_TIMESTAMP</varname> (<type>boolean</type>)</term>
869 870
      <listitem>
       <para>
871
        Prefixes each server log message with a time stamp. The default
872
        is off.
873 874 875 876 877
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
878 879 880 881
      <term><varname>SHOW_QUERY_STATS</varname> (<type>boolean</type>)</term>
      <term><varname>SHOW_PARSER_STATS</varname> (<type>boolean</type>)</term>
      <term><varname>SHOW_PLANNER_STATS</varname> (<type>boolean</type>)</term>
      <term><varname>SHOW_EXECUTOR_STATS</varname> (<type>boolean</type>)</term>
882 883
      <listitem>
       <para>
884 885 886
        For each query, write performance statistics of the respective
        module to the server log. This is a crude profiling
        instrument.
887 888 889 890 891
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
892
      <term><varname>SHOW_SOURCE_PORT</varname> (<type>boolean</type>)</term>
893 894
      <listitem>
       <para>
895 896 897 898 899
        Shows the outgoing port number of the connecting host in the
        connection log messages. You could trace back the port number
        to find out what user initiated the connection. Other than
        that it's pretty useless and therefore off by default. This
        option can only be set at server start.
900 901 902 903 904
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
905
      <term><varname>SYSLOG</varname> (<type>integer</type>)</term>
906 907
      <listitem>
       <para>
908
        <productname>Postgres</productname> allows the use of
909 910 911
        <systemitem>syslog</systemitem> for logging. If this option
        is set to 1, messages go both to <systemitem>syslog</> and the standard
        output. A setting of 2 sends output only to <systemitem>syslog</>. (Some
912
        messages will still go to the standard output/error.) The
913
        default is 0, which means <systemitem>syslog</> is off. This option must be
914 915 916
        set at server start.
       </para>
       <para>
917
        To use <systemitem>syslog</>, the build of
918 919
        <productname>Postgres</productname> must be configured with
        the <option>--enable-syslog</option> option.
920 921 922 923
       </para>
      </listitem>
     </varlistentry>

924
     <varlistentry>
925
      <term><varname>SYSLOG_FACILITY</varname> (<type>string</type>)</term>
926 927
       <listitem>
        <para>
928
          This option determines the <application>syslog</application>
929
          <quote>facility</quote> to be used when <application>syslog</application> is enabled.
930 931 932 933
          You may choose from LOCAL0, LOCAL1, LOCAL2, LOCAL3, LOCAL4,
          LOCAL5, LOCAL6, LOCAL7; the default is LOCAL0.  See also the
          documentation of your system's
          <application>syslog</application>.
934 935 936 937 938
        </para>
       </listitem>
     </varlistentry>
     
     <varlistentry>
939
      <term><varname>SYSLOG_IDENT</varname> (<type>string</type>)</term>
940 941
       <listitem>
        <para>
942
         If logging to <application>syslog</> is enabled, this option determines the
943 944 945
         program name used to identify
         <productname>PostgreSQL</productname> messages in
         <application>syslog</application> log messages.  The default
946
         is <literal>postgres</literal>.
947 948 949 950
        </para>
       </listitem>
      </varlistentry>

951
     <varlistentry>
952
      <term><varname>TRACE_NOTIFY</varname> (<type>boolean</type>)</term>
953 954
      <listitem>
       <para>
955 956 957
        Generates a great amount of debugging output for the
        <command>LISTEN</command> and <command>NOTIFY</command>
        commands.
958
       </para>
959 960 961 962 963
      </listitem>
     </varlistentry>
    </variablelist>
   </para>
   </sect2>
964

965 966
   <sect2 id="runtime-config-general">
    <title>General operation</title>
967

968 969
   <para>
    <variablelist>
970
     <varlistentry>
971
      <term><varname>AUSTRALIAN_TIMEZONES</varname> (<type>bool</type>)</term>
972 973 974 975
      <listitem>
       <para>
        If set to true, <literal>CST</literal>, <literal>EST</literal>,
        and <literal>SAT</literal> are interpreted as Australian
976 977
        time zones rather than as North American Central/Eastern
        time zones and Saturday. The default is false.
978 979 980 981
       </para>
      </listitem>
     </varlistentry>

982
     <varlistentry>
983 984 985 986 987
      <indexterm>
       <primary>deadlock</primary>
       <secondary>timeout</secondary>
      </indexterm>

988
      <term><varname>DEADLOCK_TIMEOUT</varname> (<type>integer</type>)</term>
989 990
      <listitem>
       <para>
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006
        This is the amount of time, in milliseconds, to wait on a lock
	before checking to see if there is a deadlock condition or not.
	The check for deadlock is relatively slow, so we don't want to
	run it every time we wait for a lock.  We (optimistically?)
	assume that deadlocks are not common in production applications,
	and just wait on the lock for awhile before starting to ask
	questions about whether it can ever get unlocked.
	Increasing this value reduces the amount of time wasted in
	needless deadlock checks, but slows down reporting of real deadlock
	errors.  The default is 1000 (i.e., one second), which is probably
	about the smallest value you would want in practice.  On a heavily
	loaded server you might want to raise it.  Ideally the setting
	should exceed your typical transaction time, so as to improve the
	odds that the lock will be released before the waiter decides to
	check for deadlock. 
	This option can only be set at server start.
1007 1008 1009 1010
       </para>
      </listitem>
     </varlistentry>

1011 1012 1013 1014 1015
     <varlistentry>
      <indexterm>
       <primary>transaction isolation level</primary>
      </indexterm>

1016
      <term><varname>DEFAUL_TRANSACTION_ISOLATION</varname> (<type>string</type>)</term>
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
      <listitem>
       <para>
        Each SQL transaction has an isolation level, which can be
        either <quote>read committed</quote> or
        <quote>serializable</quote>.  This parameter controls what the
        isolation level of each new transaction is set to.  The
        default is read committed.
       </para>

       <para>
        Consult the <citetitle>PostgreSQL User's Guide</citetitle> and
        the command <command>SET TRANSACTION</command> for more
        information.
       </para>
      </listitem>
     </varlistentry>

1034
     <varlistentry>
1035
      <term><varname>DYNAMIC_LIBRARY_PATH</varname> (<type>string</type>)</term>
1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076
      <listitem>
       <para>
        If a dynamically loadable module needs to be opened and the
        specified name does not have a directory component (i.e., the
        name does not contain a slash), the system will search this
        path for the specified file.  (The name that is used is the
        name specified in the <command>CREATE FUNCTION</command> or
        <command>LOAD</command> command.)
       </para>

       <para>
        The value for dynamic_library_path has to be a colon-separated
        list of absolute directory names.  If a directory name starts
        with the special value <literal>$libdir</literal>, the
        compiled-in PostgreSQL library directory, which is where the
        modules provided by the PostgreSQL distribution are installed,
        is substituted.  An example value:
        <informalexample>
<programlisting>
dynamic_library_path = '/usr/local/lib:/home/my_project/lib:$libdir:$libdir/contrib'
</programlisting>
        </informalexample>
       </para>

       <para>
        The default value for this parameter is
        <literal>$libdir</literal>.  If the value is set to the empty
        string, the automatic path search is turned off.
       </para>

       <para>
        This parameter can be changed at run time by superusers, but
        note that a setting done that way will only persist till the
        end of the client connection, so this method should be
        reserved for development purposes.  The recommended way to set
        this parameter is in the <filename>postgresql.conf</filename>
        configuration file.
       </para>
      </listitem>
     </varlistentry>

1077
     <varlistentry>
1078 1079 1080 1081
      <indexterm>
       <primary>fsync</primary>
      </indexterm>

1082
      <term><varname>FSYNC</varname> (<type>boolean</type>)</term>
1083 1084
      <listitem>
       <para>
1085
        If this option is on, the <productname>Postgres</> backend
1086 1087
        will use the <function>fsync()</> system call in several
        places to make sure that updates are physically written to
1088
        disk and do not hang around in the kernel buffer cache. This
1089 1090 1091
        increases the chance by a large amount that a database
        installation will still be usable after an operating system or
        hardware crash.  (Crashes of the database server itself do
1092 1093 1094 1095
        <emphasis>not</> affect this consideration.)
       </para>

       <para>
1096 1097
        However, this operation slows down <productname>Postgres</>,
	because at all those points it has
1098 1099 1100
        to block and wait for the operating system to flush the
        buffers. Without <function>fsync</>, the operating system is
        allowed to do its best in buffering, sorting, and delaying
1101
        writes, which can make for a considerable performance
1102 1103 1104
        increase. However, if the system crashes, the results of the
	last few committed transactions may be lost in part or whole;
	in the worst case, unrecoverable data corruption may occur.
1105 1106 1107 1108 1109 1110 1111 1112 1113
       </para>

       <para>
        This option is the subject of an eternal debate in the
        <productname>Postgres</> user and developer communities. Some
        always leave it off, some turn it off only for bulk loads,
        where there is a clear restart point if something goes wrong,
        some leave it on just to be on the safe side. Because it is
        the safe side, on is also the default. If you trust your
1114
        operating system, your hardware, and your utility company (or
1115
	better your UPS), you might want to disable <varname>fsync</varname>.
1116 1117 1118 1119 1120 1121 1122 1123
       </para>

       <para>
        It should be noted that the performance penalty from doing
	fsyncs is considerably less in <productname>Postgres</> version
	7.1 than it was in prior releases.  If you previously suppressed
	fsyncs because of performance problems, you may wish to reconsider
	your choice.
1124
       </para>
Tom Lane's avatar
Tom Lane committed
1125 1126 1127 1128 1129

       <para>
	This option can only be set at server start or in the
	<filename>postgresql.conf</filename> file.
       </para>
1130 1131
      </listitem>
     </varlistentry>
1132

1133
     <varlistentry>
1134
      <term><varname>KRB_SERVER_KEYFILE</varname> (<type>string</type>)</term>
1135 1136 1137 1138 1139 1140 1141 1142
      <listitem>
       <para>
        Sets the location of the Kerberos server key file. See
        <xref linkend="kerberos-auth"> for details.
       </para>
      </listitem>
     </varlistentry>

1143
     <varlistentry>
1144
      <term><varname>MAX_CONNECTIONS</varname> (<type>integer</type>)</term>
1145 1146 1147
      <listitem>
       <para>
        Determines how many concurrent connections the database server
1148 1149
        will allow. The default is 32 (unless altered while building
	the server). This parameter can only be set at server start.
1150 1151 1152
       </para>
      </listitem>
     </varlistentry>
1153

1154
     <varlistentry>
1155
      <term><varname>MAX_EXPR_DEPTH</varname> (<type>integer</type>)</term>
1156 1157 1158 1159 1160 1161 1162 1163
      <listitem>
       <para>
        Sets the maximum expression nesting depth that the parser will
	accept. The default value is high enough for any normal query,
	but you can raise it if you need to. (But if you raise it too
	high, you run the risk of backend crashes due to stack
	overflow.)
       </para>
1164 1165 1166 1167
      </listitem>
     </varlistentry>

     <varlistentry>
1168
      <term><varname>MAX_FSM_RELATIONS</varname> (<type>integer</type>)</term>
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
      <listitem>
       <para>
        Sets the maximum number of relations (tables) for which free space
	will be tracked in the shared free-space map.
	The default is 100.  This option can only be set at server start.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1179
      <term><varname>MAX_FSM_PAGES</varname> (<type>integer</type>)</term>
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
      <listitem>
       <para>
        Sets the maximum number of disk pages for which free space
	will be tracked in the shared free-space map.
	The default is 10000.  This option can only be set at server start.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1190
      <term><varname>MAX_LOCKS_PER_TRANSACTION</varname> (<type>integer</type>)</term>
1191 1192 1193
      <listitem>
       <para>
        The shared lock table is sized on the assumption that at most
1194
	<varname>max_locks_per_transaction</> * <varname>max_connections</varname> distinct objects will need
1195 1196 1197 1198 1199 1200 1201 1202 1203
	to be locked at any one time.  The default, 64, has historically
	proven sufficient, but you might need to raise this value if you
	have clients that touch many different tables in a single transaction.
	This option can only be set at server start.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1204
      <term><varname>PORT</varname> (<type>integer</type>)</term>
1205 1206
      <listitem>
       <para>
1207 1208
        The TCP port the server listens on; 5432 by default. This
        option can only be set at server start.
1209 1210 1211 1212 1213
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1214
      <term><varname>SHARED_BUFFERS</varname> (<type>integer</type>)</term>
1215 1216
      <listitem>
       <para>
1217 1218 1219
        Sets the number of shared memory buffers the database server
        will use. The default is 64. Each buffer is typically 8192
        bytes. This option can only be set at server start.
1220
       </para>
1221 1222 1223 1224
      </listitem>
     </varlistentry>

     <varlistentry>
1225
      <term><varname>SILENT_MODE</varname> (<type>bool</type>)</term>
1226 1227
      <listitem>
       <para>
1228
        Runs postmaster silently. If this option is set, postmaster
1229
        will automatically run in background and any controlling ttys
1230 1231 1232
        are disassociated, thus no messages are written to standard output or
        standard error (same effect as postmaster's -S option). Unless some
        logging system such as <application>syslog</> is enabled, using this option is
1233
        discouraged since it makes it impossible to see error
1234 1235
        messages.
       </para>
1236 1237 1238
      </listitem>
     </varlistentry>

1239
     <varlistentry>
1240
      <term><varname>SORT_MEM</varname> (<type>integer</type>)</term>
1241 1242 1243 1244 1245 1246 1247 1248
      <listitem>
       <para>
	Specifies the amount of memory to be used by internal sorts
	and hashes before resorting to temporary disk files. The value
	is specified in kilobytes, and defaults to 512 kilobytes. Note
	that for a complex query, several sorts and/or hashes might be
	running in parallel, and each one will be allowed to use as
	much memory as this value specifies before it starts to put
1249 1250
	data into temporary files.  And don't forget that each running
	backend could be doing one or more sorts.  So the total memory
1251
	space needed could be many times the value of <varname>SORT_MEM</varname>.
1252 1253 1254
       </para>
      </listitem>
     </varlistentry>
1255 1256

     <varlistentry>
1257
      <term><varname>SQL_INHERITANCE</varname> (<type>bool</type>)</term>
1258 1259 1260 1261 1262
      <listitem>
       <para>
        This controls the inheritance semantics, in particular whether
        subtables are included into the consideration of various
        commands by default. This was not the case in versions prior
1263
        to 7.1. If you need the old behavior you can set this
1264 1265 1266 1267 1268
        variable to off, but in the long run you are encouraged to
        change your applications to use the <literal>ONLY</literal>
        keyword to exclude subtables. See the SQL language reference
        and the <citetitle>User's Guide</citetitle> for more
        information about inheritance.
1269 1270 1271
       </para>
      </listitem>
     </varlistentry>
1272 1273

     <varlistentry>
1274 1275 1276 1277
      <indexterm>
       <primary>SSL</primary>
      </indexterm>

1278
      <term><varname>SSL</varname> (<type>boolean</type>)</term>
1279 1280 1281
      <listitem>
       <para>
        Enables <acronym>SSL</> connections. Please read
1282
        <xref linkend="ssl-tcp"> before using this. The default
1283 1284 1285 1286 1287 1288
        is off.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1289
      <term><varname>TCPIP_SOCKET</varname> (<type>boolean</type>)</term>
1290 1291 1292 1293 1294 1295 1296 1297 1298
      <listitem>
       <para>
        If this is true, then the server will accept TCP/IP
        connections. Otherwise only local Unix domain socket
        connections are accepted. It is off by default. This option
        can only be set at server start.
       </para>
      </listitem>
     </varlistentry>
1299

1300
     <varlistentry>
1301
      <term><varname>UNIX_SOCKET_DIRECTORY</varname> (<type>string</type>)</term>
1302 1303
      <listitem>
       <para>
1304
	Specifies the directory of the Unix-domain socket on which the
1305
	<application>postmaster</application> is to listen for
1306 1307
	connections from client applications.  The default is normally
	<filename>/tmp</filename>, but can be changed at build time.
1308 1309 1310 1311
       </para>
      </listitem>
     </varlistentry>

1312
     <varlistentry>
1313
      <term><varname>UNIX_SOCKET_GROUP</varname> (<type>string</type>)</term>
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
      <listitem>
       <para>
        Sets the group owner of the Unix domain socket.  (The owning
        user of the socket is always the user that starts the
        postmaster.)  In combination with the option
        <option>UNIX_SOCKET_PERMISSIONS</option> this can be used as
        an additional access control mechanism for this socket type.
        By default this is the empty string, which uses the default
        group for the current user.  This option can only be set at
        server start.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1329
      <term><varname>UNIX_SOCKET_PERMISSIONS</varname> (<type>integer</type>)</term>
1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
      <listitem>
       <para>
        Sets the access permissions of the Unix domain socket.  Unix
        domain sockets use the usual Unix file system permission set.
        The option value is expected to be an numeric mode
        specification in the form accepted by the
        <function>chmod</function> and <function>umask</function>
        system calls.  (To use the customary octal format the number
        must start with a <literal>0</literal> (zero).)
       </para>

       <para>
        The default permissions are <literal>0777</literal>, meaning
        anyone can connect.  Reasonable alternatives would be
        <literal>0770</literal> (only user and group, see also under
        <option>UNIX_SOCKET_GROUP</option>) and
        <literal>0700</literal> (only user).  (Note that actually for
        a Unix socket, only write permission matters and there is no
        point in setting or revoking read or execute permissions.)
       </para>

       <para>
        This access control mechanism is independent from the one
        described in <xref linkend="client-authentication">.
       </para>

       <para>
        This option can only be set at server start.
       </para>
      </listitem>
     </varlistentry>

1362
     <varlistentry>
1363
      <term><varname>VIRTUAL_HOST</varname> (<type>string</type>)</term>
1364 1365
      <listitem>
       <para>
1366
	Specifies the TCP/IP host name or address on which the
1367
	<application>postmaster</application> is to listen for
1368
	connections from client applications.  Defaults to
1369
	listening on all configured addresses (including <systemitem class="systemname">localhost</>).
1370 1371 1372 1373
       </para>
      </listitem>
     </varlistentry>

1374
    </variablelist>
1375
   </para>
1376
   </sect2>
1377

1378 1379 1380 1381 1382 1383 1384 1385
   <sect2 id="runtime-config-wal">
    <title>WAL</title>

   <para>
    See also <xref linkend="wal-configuration"> for details on WAL
    tuning.

    <variablelist>
Tom Lane's avatar
Tom Lane committed
1386
     <varlistentry>
1387
      <term><varname>CHECKPOINT_SEGMENTS</varname> (<type>integer</type>)</term>
Tom Lane's avatar
Tom Lane committed
1388 1389
      <listitem>
       <para>
1390
        Maximum distance between automatic WAL checkpoints, in log file
Tom Lane's avatar
Tom Lane committed
1391 1392 1393 1394 1395 1396 1397
	segments (each segment is normally 16 megabytes).
	This option can only be set at server start or in the
	<filename>postgresql.conf</filename> file.
       </para>
      </listitem>
     </varlistentry>

1398
     <varlistentry>
1399
      <term><varname>CHECKPOINT_TIMEOUT</varname> (<type>integer</type>)</term>
1400 1401
      <listitem>
       <para>
Tom Lane's avatar
Tom Lane committed
1402 1403 1404
        Maximum time between automatic WAL checkpoints, in seconds.
	This option can only be set at server start or in the
	<filename>postgresql.conf</filename> file.
1405 1406 1407 1408
       </para>
      </listitem>
     </varlistentry>

1409
     <varlistentry>
1410
      <term><varname>COMMIT_DELAY</varname> (<type>integer</type>)</term>
1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426
      <listitem>
       <para>
        Time delay between writing a commit record to the WAL buffer and
	flushing the buffer out to disk, in microseconds.  A nonzero delay
	allows multiple transactions to be committed with only one fsync,
	if system load is high enough that additional transactions become
	ready to commit within the given interval.  But the delay is just
	wasted time if no other transactions become ready to commit.
	Therefore, the delay is only performed if at least COMMIT_SIBLINGS
	other transactions are active at the instant that a backend has
	written its commit record.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1427
      <term><varname>COMMIT_SIBLINGS</varname> (<type>integer</type>)</term>
1428 1429 1430 1431 1432 1433 1434 1435 1436 1437
      <listitem>
       <para>
        Minimum number of concurrent open transactions to require before
	performing the COMMIT_DELAY delay.  A larger value makes it more
	probable that at least one other transaction will become ready to
	commit during the delay interval.
       </para>
      </listitem>
     </varlistentry>

1438
     <varlistentry>
1439
      <term><varname>WAL_BUFFERS</varname> (<type>integer</type>)</term>
1440 1441
      <listitem>
       <para>
1442 1443
        Number of disk-page buffers in shared memory for WAL log.
	This option can only be set at server start.
1444 1445 1446 1447 1448
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1449
      <term><varname>WAL_DEBUG</varname> (<type>integer</type>)</term>
1450 1451 1452 1453 1454 1455 1456 1457 1458
      <listitem>
       <para>
        If non-zero, turn on WAL-related debugging output on standard
        error.
       </para>
      </listitem>
     </varlistentry>

     <varlistentry>
1459
      <term><varname>WAL_FILES</varname> (<type>integer</type>)</term>
1460 1461 1462
      <listitem>
       <para>
        Number of log files that are created in advance at checkpoint
Tom Lane's avatar
Tom Lane committed
1463 1464
        time.  This option can only be set at server start or in the
	<filename>postgresql.conf</filename> file.
1465 1466 1467
       </para>
      </listitem>
     </varlistentry>
1468 1469

     <varlistentry>
1470
      <term><varname>WAL_SYNC_METHOD</varname> (<type>string</type>)</term>
1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
      <listitem>
       <para>
        Method used for forcing WAL updates out to disk.  Possible
	values are
	<literal>FSYNC</> (call fsync() at each commit),
	<literal>FDATASYNC</> (call fdatasync() at each commit),
	<literal>OPEN_SYNC</> (write WAL files with open() option O_SYNC), or
	<literal>OPEN_DATASYNC</> (write WAL files with open() option O_DSYNC).
	Not all of these choices are available on all platforms.
        This option can only be set at server start or in the
	<filename>postgresql.conf</filename> file.
       </para>
      </listitem>
     </varlistentry>
1485 1486 1487 1488 1489
    </variablelist>
    </para>
   </sect2>


1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
   <sect2 id="runtime-config-short">
    <title>Short options</title>
   <para>
    For convenience there are also single letter option switches
    available for many parameters. They are described in the following
    table.

    <table>
     <title>Short option key</title>
     <tgroup cols="3">
      <colspec colnum="3" align="center">
      <thead>
       <row>
        <entry>Short option</entry>
        <entry>Equivalent</entry>
        <entry>Remark</entry>
       </row>
      </thead>
      <tbody>
       <row>
1510 1511
        <entry><option>-B <replaceable>x</replaceable></option></entry>
        <entry><literal>shared_buffers = <replaceable>x</replaceable></></entry>
1512 1513 1514
        <entry></entry>
       </row>
       <row>
1515 1516
        <entry><option>-d <replaceable>x</replaceable></option></entry>
        <entry><literal>debug_level = <replaceable>x</replaceable></></entry>
1517 1518 1519
        <entry></entry>
       </row>
       <row>
1520 1521
        <entry><option>-F</option></entry>
        <entry><literal>fsync = off</></entry>
1522 1523
        <entry></entry>
       </row>
1524
       <row>
1525 1526
        <entry><option>-h <replaceable>x</replaceable></option></entry>
        <entry><literal>virtual_host = <replaceable>x</replaceable></></entry>
1527 1528
        <entry></entry>
       </row>
1529
       <row>
1530 1531
        <entry><option>-i</option></entry>
        <entry><literal>tcpip_socket = on</></entry>
1532
        <entry></entry>
1533 1534
       </row>
       <row>
1535 1536
        <entry><option>-k <replaceable>x</replaceable></option></entry>
        <entry><literal>unix_socket_directory = <replaceable>x</replaceable></></entry>
1537
        <entry></entry>
1538
       </row>
1539
       <row>
1540 1541
        <entry><option>-l</option></entry>
        <entry><literal>ssl = on</></entry>
1542 1543
        <entry></entry>
       </row>
1544
       <row>
1545 1546
        <entry><option>-N <replaceable>x</replaceable></option></entry>
        <entry><literal>max_connections = <replaceable>x</replaceable></></entry>
1547 1548 1549
        <entry></entry>
       </row>
       <row>
1550 1551
        <entry><option>-p <replaceable>x</replaceable></option></entry>
        <entry><literal>port = <replaceable>x</replaceable></></entry>
1552 1553 1554 1555
        <entry></entry>
       </row>

       <row>
1556 1557 1558 1559
        <entry><option>-fi</option>, <option>-fh</option>, <option>-fm</option>, <option>-fn</option>, <option>-fs</option>, <option>-ft</option></entry>
        <entry><literal>enable_indexscan=off</>, <literal>enable_hashjoin=off</>,
        <literal>enable_mergejoin=off</>, <literal>enable_nestloop=off</>, <literal>enable_seqscan=off</>,
        <literal>enable_tidscan=off</></entry>
1560 1561 1562
        <entry>*</entry>
       </row>
       <row>
1563 1564
        <entry><option>-S <replaceable>x</replaceable></option></entry>
        <entry><literal>sort_mem = <replaceable>x</replaceable></></entry>
1565 1566 1567
        <entry>*</entry>
       </row>
       <row>
1568 1569
        <entry><option>-s</option></entry>
        <entry><literal>show_query_stats = on</></entry>
1570 1571 1572
        <entry>*</entry>
       </row>
       <row>
1573 1574
        <entry><option>-tpa</option>, <option>-tpl</option>, <option>-te</option></entry>
        <entry><literal>show_parser_stats=on</>, <literal>show_planner_stats=on</>, <literal>show_executor_stats=on</></entry>
1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592
        <entry>*</entry>
       </row>
      </tbody>
     </tgroup>
    </table>
    For historical reasons, options marked <quote>*</quote> must be
    passed to the individual backend process via the
    <option>-o</option> postmaster option, for example,
<screen>
&gt; <userinput>postmaster -o '-S 1024 -s'</userinput>
</screen>
    or via <envar>PGOPTIONS</envar> from the client side, as explained
    above.
   </para>

   </sect2>
 </sect1>

1593

1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
 <sect1 id="kernel-resources">
  <title>Managing Kernel Resources</title>

  <para>
   A large <productname>Postgres</> installation can quickly hit
   various operating system resource limits. (On some systems, the
   factory defaults are so low that you don't even need a really
   <quote>large</> installation.) If you have encountered this kind of
   problem then keep reading.
  </para>

  <sect2 id="sysvipc">
   <title>Shared Memory and Semaphores</title>

1608 1609 1610 1611 1612 1613 1614 1615
   <indexterm zone="sysvipc">
    <primary>shared memory</primary>
   </indexterm>

   <indexterm zone="sysvipc">
    <primary>semaphores</primary>
   </indexterm>

1616 1617
   <para>
    Shared memory and semaphores are collectively referred to as
1618
    <quote><systemitem class="osname">System V</> <acronym>IPC</></quote> (together with message queues, which are
1619 1620 1621
    not relevant for <productname>Postgres</>). Almost all modern
    operating systems provide these features, but not all of them have
    them turned on or sufficiently sized by default, especially
1622
    systems with BSD heritage. (For the <systemitem class="osname">QNX</> and <systemitem class="osname">BeOS</> ports,
1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
    <productname>Postgres</> provides its own replacement
    implementation of these facilities.)
   </para>

   <para>
    The complete lack of these facilities is usually manifested by an
    <errorname>Illegal system call</> error upon postmaster start. In
    that case there's nothing left to do but to reconfigure your
    kernel -- <productname>Postgres</> won't work without them.
   </para>

   <para>
    When <productname>Postgres</> exceeds one of the various hard
1636
    limits of the <acronym>IPC</> resources then the postmaster will refuse to
1637 1638
    start up and should leave a marginally instructive error message
    about which problem was encountered and what needs to be done
1639 1640
    about it.  (See also <xref linkend="postmaster-start-failures">.)
    The relevant kernel parameters are named
Tom Lane's avatar
Tom Lane committed
1641
    consistently across different systems; <xref
1642 1643
    linkend="sysvipc-parameters"> gives an overview. The methods to
    set them, however, vary; suggestions for some platforms are given
1644
    below. Be warned that it is often necessary to reboot your
Tom Lane's avatar
Tom Lane committed
1645
    machine at least, possibly even recompile the kernel, to change these
1646 1647 1648 1649 1650
    settings.
   </para>


   <table id="sysvipc-parameters">
1651
    <title><systemitem class="osname">System V</> <acronym>IPC</> parameters</>
1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665

    <tgroup cols="3">
     <thead>
      <row>
       <entry>Name</>
       <entry>Description</>
       <entry>Reasonable values</>
      </row>
     </thead>

     <tbody>
      <row>
       <entry><varname>SHMMAX</></>
       <entry>Maximum size of shared memory segment (bytes)</>
1666
       <entry>250 kB + 8.2kB * <varname>shared_buffers</> + 14.2kB * <varname>max_connections</> or infinity</entry>
1667 1668 1669 1670 1671
      </row>

      <row>
       <entry><varname>SHMMIN</></>
       <entry>Minimum size of shared memory segment (bytes)</>
Bruce Momjian's avatar
Bruce Momjian committed
1672
       <entry>1</>
1673 1674
      </row>

1675 1676 1677
      <row>
       <entry><varname>SHMALL</></>
       <entry>Total amount of shared memory available (bytes or pages)</>
1678
       <entry>if bytes, same as <varname>SHMMAX</varname>; if pages, <literal>ceil(SHMMAX/PAGE_SIZE)</literal></>
1679 1680
      </row>

1681 1682 1683
      <row>
       <entry><varname>SHMSEG</></>
       <entry>Maximum number of shared memory segments per process</>
1684
       <entry>only 1 segment is needed, but the default is much higher</>
1685 1686 1687 1688 1689
      </row>

       <row>
        <entry><varname>SHMMNI</></>
        <entry>Maximum number of shared memory segments system-wide</>
Bruce Momjian's avatar
Bruce Momjian committed
1690
        <entry>like <varname>SHMSEG</> plus room for other applications</>
1691 1692 1693 1694 1695
       </row>

       <row>
        <entry><varname>SEMMNI</></>
        <entry>Maximum number of semaphore identifiers (i.e., sets)</>
1696
        <entry><literal>&gt;= ceil(max_connections / 16)</literal></>
1697 1698 1699 1700 1701
       </row>

       <row>
        <entry><varname>SEMMNS</></>
        <entry>Maximum number of semaphores system-wide</>
1702
        <entry><literal>ceil(max_connections / 16) * 17</literal> + room for other applications</>
1703 1704 1705 1706 1707
       </row>

       <row>
        <entry><varname>SEMMSL</></>
        <entry>Maximum number of semaphores per set</>
1708
        <entry>&gt;= 17</>
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728
       </row>

       <row>
        <entry><varname>SEMMAP</></>
        <entry>Number of entries in semaphore map</>
        <entry>see text</>
       </row>

       <row>
        <entry><varname>SEMVMX</></>
        <entry>Maximum value of semaphore</>
        <entry>&gt;= 255  (The default is often 32767, don't change unless asked to.)</>
       </row>

     </tbody>
    </tgroup>
   </table>


   <para>
1729
    <indexterm><primary>SHMMAX</primary></indexterm>
1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
    The most important shared memory parameter is <varname>SHMMAX</>,
    the maximum size, in bytes, that a shared memory segment can have.
    If you get an error message from <function>shmget</> along the
    lines of <errorname>Invalid argument</> then it is possible that
    this limit has been exceeded. The size of the required shared
    memory segments varies both with the number of requested buffers
    (<option>-B</> option) and the number of allowed connections
    (<option>-N</> option), although the former is the dominant item.
    (You can therefore, as a temporary solution, lower these settings
    to get rid of the failures.) As a rough approximation you can
    estimate the required segment size as the number of buffers times
    the block size (8192 kB by default) plus ample overhead (at least
    half a megabyte). Any error message you might get will contain the
1743
    size of the failed allocation request.
1744 1745 1746 1747
   </para>

   <para>
    Less likely to cause problems is the minimum size for shared
1748 1749 1750 1751 1752 1753 1754
    memory segments (<varname>SHMMIN</>), which should be at most
    somewhere around 256 kB for <productname>Postgres</> (it is
    usually just 1). The maximum number of segments system-wide
    (<varname>SHMMNI</>) or per-process (<varname>SHMSEG</>) should
    not cause a problem unless your system has them set to zero. Some
    systems also have a limit on the total amount of shared memory in
    the system; see the platform-specific instructions below.
1755 1756 1757 1758
   </para>

   <para>
    <productname>Postgres</> uses one semaphore per allowed connection
1759 1760
    (<option>-N</> option), in sets of 16.  Each such set will also
    contain a 17th semaphore which contains a <quote>magic
1761
    number</quote>, to detect collision with semaphore sets used by
1762 1763 1764 1765
    other applications. The maximum number of semaphores in the system
    is set by <varname>SEMMNS</>, which consequently must be at least
    as high as the connection setting plus one extra for each 16
    allowed connections (see the formula in <xref
1766
    linkend="sysvipc-parameters">).  The parameter <varname>SEMMNI</>
1767 1768 1769 1770 1771 1772
    determines the limit on the number of semaphore sets that can
    exist on the system at one time.  Hence this parameter must be at
    least <literal>ceil(max_connections / 16)</>. Lowering the number
    of allowed connections is a temporary workaround for failures,
    which are usually confusingly worded <quote><errorname>No space
    left on device</></>, from the function <function>semget()</>.
1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
   </para>

   <para>
    In some cases it might also turn out to be necessary to increase
    <varname>SEMMAP</> to be at least on the order of
    <varname>SEMMNS</>. This parameter defines the size of the
    semaphore resource map, in which each contiguous block of available
    semaphores needs an entry. When a semaphore set is freed it is
    either added to an existing entry that is adjacent to the freed
    block or it is registered under a new map entry. If the map is
Tom Lane's avatar
Tom Lane committed
1783
    full, the freed semaphores get lost (until reboot). Fragmentation
1784 1785 1786 1787 1788 1789
    of the semaphore space could therefore over time lead to less
    available semaphores than there should be.
   </para>

   <para>
    The <varname>SEMMSL</> parameter, which determines how many
1790
    semaphores can be in a set, must be at least 17 for
1791 1792 1793 1794
    <productname>Postgres</>.
   </para>

   <para>
Tom Lane's avatar
Tom Lane committed
1795 1796 1797
    Various other settings related to <quote>semaphore undo</>, such as
    <varname>SEMMNU</> and <varname>SEMUME</>, are not of concern
    for <productname>Postgres</>.
1798 1799 1800 1801 1802 1803 1804
   </para>


   <para>
    <variablelist>

     <varlistentry>
1805
      <term><systemitem class="osname">BSD/OS</></term>
1806 1807 1808 1809 1810 1811
      <listitem>
       <formalpara>
        <title>Shared Memory</>
        <para>
         By default, only 4 MB of shared memory is supported. Keep in
         mind that shared memory is not pageable; it is locked in RAM.
1812 1813
	 
	 To increase the number of shared buffers supported by the
1814
	 postmaster, add the following to your kernel configuration file. A
1815 1816 1817
	 <varname>SHMALL</> value of 1024 represents 4MB of shared
	 memory. The following increases the maximum shared memory area
	 to 32 MB:
1818
<programlisting>
1819
options "SHMALL=8192"
Bruce Momjian's avatar
Bruce Momjian committed
1820
options "SHMMAX=\(SHMALL*PAGE_SIZE\)"
1821 1822 1823 1824 1825
</programlisting>
        </para>
       </formalpara>

       <para>
1826 1827 1828 1829
        For those running 4.1 or later, just make the above changes,
        recompile the kernel, and reboot. For those running earlier
        releases, use <application>bpatch</> to find the
        <varname>sysptsize</> value in the current kernel. This is
1830
        computed dynamically at boot time.
1831 1832 1833 1834
<screen>
$ <userinput>bpatch -r sysptsize</>
<computeroutput>0x9 = 9</>
</screen>
1835
        Next, add <varname>SYSPTSIZE</> as a hard-coded value in the
1836
        kernel configuration file. Increase the value you found using
1837
        <application>bpatch</>. Add 1 for every additional 4 MB of
1838 1839
        shared memory you desire.
<programlisting>
1840
options "SYSPTSIZE=16"
1841
</programlisting>
1842
        <varname>sysptsize</> cannot be changed by <command>sysctl</command>.
1843 1844 1845 1846 1847 1848
       </para>

       <formalpara>
        <title>Semaphores</>
        <para>
         You may need to increase the number of semaphores. By
1849 1850
         default, <productname>Postgres</> allocates 34 semaphores,
	 which is over half the default system total of 60.
1851 1852 1853
        </para>
       </formalpara>

Bruce Momjian's avatar
Bruce Momjian committed
1854
        <para>
1855
        Set the values you want in your kernel configuration file, e.g.:
1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
<programlisting>
options "SEMMNI=40"
options "SEMMNS=240"
options "SEMUME=40"
options "SEMMNU=120"
</programlisting>
       </para>
      </listitem>
     </varlistentry>


     <varlistentry>
1868 1869 1870
      <term><systemitem class="osname">FreeBSD</></term>
      <term><systemitem class="osname">NetBSD</></term>
      <term><systemitem class="osname">OpenBSD</></term>
1871 1872 1873 1874 1875 1876
      <listitem>
       <para>
        The options <varname>SYSVSHM</> and <varname>SYSVSEM</> need
        to be enabled when the kernel is compiled. (They are by
        default.) The maximum size of shared memory is determined by
        the option <varname>SHMMAXPGS</> (in pages). The following
Tom Lane's avatar
Tom Lane committed
1877
        shows an example of how to set the various parameters:
1878
<programlisting>
1879 1880 1881 1882 1883 1884 1885 1886
options         SYSVSHM
options         SHMMAXPGS=4096
options         SHMSEG=256

options         SYSVSEM
options         SEMMNI=256
options         SEMMNS=512
options         SEMMNU=256
1887 1888
options         SEMMAP=256
</programlisting>
1889 1890
        (On <systemitem class="osname">NetBSD</> and <systemitem
        class="osname">OpenBSD</> the key word is actually
1891
        <literal>option</literal> singular.)
1892 1893 1894 1895 1896 1897
       </para>
      </listitem>
     </varlistentry>


     <varlistentry>
1898
      <term><systemitem class="osname">HP-UX</></term>
1899 1900 1901
      <listitem>
       <para>
        The default settings tend to suffice for normal installations.
1902
        On <productname>HP-UX</> 10, the factory default for
1903 1904 1905 1906
        <varname>SEMMNS</> is 128, which might be too low for larger
        database sites.
       </para>
       <para>
1907
        <acronym>IPC</> parameters can be set in the <application>System
1908 1909 1910 1911 1912 1913 1914 1915 1916 1917
        Administration Manager</> (<acronym>SAM</>) under
        <menuchoice><guimenu>Kernel
        Configuration</><guimenuitem>Configurable Parameters</></>.
        Hit <guibutton>Create A New Kernel</> when you're done.
       </para>
      </listitem>
     </varlistentry>


     <varlistentry>
1918
      <term><systemitem class="osname">Linux</></term>
1919 1920
      <listitem>
       <para>
1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932
        The default shared memory limit (both
        <varname>SHMMAX</varname> and <varname>SHMALL</varname>) is 32
        MB in 2.2 kernels, but it can be changed in the
        <filename>proc</filename> file system (without reboot).  For
        example, to allow 128 MB:
<screen>
<prompt>$</prompt> <userinput>echo 134217728 >/proc/sys/kernel/shmall</userinput>
<prompt>$</prompt> <userinput>echo 134217728 >/proc/sys/kernel/shmmax</userinput>
</screen>
        You could put these commands into a script run at boot-time.
       </para>

1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948
       <para>
        Alternatively, you can use
        <citerefentry><refentrytitle>sysctl</refentrytitle>
        <manvolnum>8</manvolnum></citerefentry>, if available, to
        control these parameters.  Look for a file called
        <filename>/etc/sysctl.conf</filename> and add lines like the
        following to it:
<programlisting>
kernel.shmall = 134217728
kernel.shmmax = 134217728
</programlisting>
        This file is usually processed at boot time, but
        <application>sysctl</application> can also be called
        explicitly later.
       </para>

1949 1950 1951
       <para>
        Other parameters are sufficiently sized for any application.
        If you want to see for yourself look into
1952
        <filename>/usr/src/linux/include/asm-<replaceable>xxx</>/shmparam.h</>
1953
        and <filename>/usr/src/linux/include/linux/sem.h</>.
1954 1955 1956 1957 1958 1959
       </para>
      </listitem>
     </varlistentry>


     <varlistentry>
1960
      <term><systemitem class="osname">SCO OpenServer</></term>
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986
      <listitem>
       <para>
        In the default configuration, only 512 kB of shared memory per
        segment is allowed, which is about enough for <option>-B 24 -N
        12</>. To increase the setting, first change the directory to
        <filename>/etc/conf/cf.d</>. To display the current value of
        <varname>SHMMAX</>, in bytes, run
<programlisting>
./configure -y SHMMAX
</programlisting>
        To set a new value for <varname>SHMMAX</>, run:
<programlisting>
./configure SHMMAX=<replaceable>value</>
</programlisting>
        where <replaceable>value</> is the new value you want to use
        (in bytes). After setting <varname>SHMMAX</>, rebuild the kernel
<programlisting>
./link_unix
</programlisting>
        and reboot.
       </para>
      </listitem>
     </varlistentry>


     <varlistentry>
1987
      <term><systemitem class="osname">Solaris</></term>
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
      <listitem>
       <para>
        At least in version 2.6, the maximum size of a shared memory
        segment is set too low for <productname>Postgres</>. The
        relevant settings can be changed in <filename>/etc/system</>,
        for example:
<programlisting>
set shmsys:shminfo_shmmax=0x2000000
set shmsys:shminfo_shmmin=1
set shmsys:shminfo_shmmni=256
set shmsys:shminfo_shmseg=256

set semsys:seminfo_semmap=256
set semsys:seminfo_semmni=512
set semsys:seminfo_semmns=512
set semsys:seminfo_semmsl=32
</programlisting>
        You need to reboot to make the changes effective.
       </para>

       <para>
        See also <ulink
        url="http://www.sunworld.com/swol-09-1997/swol-09-insidesolaris.html">http://www.sunworld.com/swol-09-1997/swol-09-insidesolaris.html</>
        for information on shared memory under
        <productname>Solaris</>.
       </para>
      </listitem>
     </varlistentry>


     <varlistentry>
2019
      <term><systemitem class="osname">UnixWare</></term>
2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
      <listitem>
       <para>
        On <productname>UnixWare</> 7, the maximum size for shared
        memory segments is 512 kB in the default configuration. This
        is enough for about <option>-B 24 -N 12</>. To display the
        current value of <varname>SHMMAX</>, run
<programlisting>
/etc/conf/bin/idtune -g SHMMAX
</programlisting>
        which displays the current, default, minimum, and maximum
        values, in bytes. To set a new value for <varname>SHMMAX</>,
        run:
<programlisting>
/etc/conf/bin/idtune SHMMAX <replaceable>value</>
</programlisting>
        where <replaceable>value</> is the new value you want to use
        (in bytes). After setting <varname>SHMMAX</>, rebuild the
        kernel
<programlisting>
/etc/conf/bin/idbuild -B
</programlisting>
        and reboot.
       </para>
      </listitem>
     </varlistentry>

    </variablelist>

   </para>
  </sect2>

2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065

  <sect2>
   <title>Resource Limits</title>

   <para>
    Unix-like operating systems enforce various kinds of resource
    limits that might interfere with the operation of your
    <productname>Postgres</productname> server.  Of importance are
    especially the limits on the number of processes per user, the
    number of open files per process, and the amount of memory
    available to a process.  Each of these have a <quote>hard</quote>
    and a <quote>soft</quote> limit.  The soft limit is what actually
    counts but it can be changed by the user up to the hard limit.
    The hard limit can only be changed by the root user.  The system
    call <function>setrlimit</function> is responsible for setting
2066
    these parameters.  The shell's built-in command
2067
    <command>ulimit</command> (Bourne shells) or
2068
    <command>limit</command> (<application>csh</>) is used to control the resource
2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091
    limits from the command line.  On BSD-derived systems the file
    <filename>/etc/login.conf</filename> controls what values the
    various resource limits are set to upon login.  See
    <citerefentry><refentrytitle>login.conf</refentrytitle>
    <manvolnum>5</manvolnum></citerefentry> for details.  The relevant
    parameters are <varname>maxproc</varname>,
    <varname>openfiles</varname>, and <varname>datasize</varname>.
    For example:
<programlisting>
default:\
...
        :datasize-cur=256M:\
        :maxproc-cur=256:\
        :openfiles-cur=256:\
...
</programlisting>
    (<literal>-cur</literal> is the soft limit.  Append
    <literal>-max</literal> to set the hard limit.)
   </para>

   <para>
    Kernels generally also have an implementation-dependent
    system-wide limit on some resources.
2092 2093 2094
    <itemizedlist>
     <listitem>
      <para>
2095 2096 2097 2098 2099 2100 2101 2102 2103
      On <productname>Linux</productname>
      <filename>/proc/sys/fs/file-max</filename> determines the
      maximum number of files that the kernel will allocate.  It can
      be changed by writing a different number into the file or by
      adding an assignment in <filename>/etc/sysctl.conf</filename>.
      The maximum limit of files per process is fixed at the time the
      kernel is compiled; see
      <filename>/usr/src/linux/Documentation/proc.txt</filename> for
      more information.
2104 2105 2106
      </para>
     </listitem>
    </itemizedlist>
2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125
   </para>

   <para>
    The <productname>Postgres</productname> server uses one process
    per connection so you should provide for at least as many processes
    as allowed connections, in addition to what you need for the rest
    of your system.  This is usually not a problem but if you run
    several servers on one machine things might get tight.
   </para>

   <para>
    The factory default limit on open files is often set to
    <quote>socially friendly</quote> values that allow many users to
    coexist on a machine without using an inappropriate fraction of
    the system resources.  If you run many servers on a machine this
    is perhaps what you want, but on dedicated servers you may want to
    raise this limit.
   </para>
  </sect2>
2126 2127 2128 2129

 </sect1>


2130 2131 2132 2133 2134 2135 2136 2137 2138
 <sect1 id="postmaster-shutdown">
  <title>Shutting down the server</title>

  <para>
   Depending on your needs, there are several ways to shut down the
   database server when your work is done. The differentiation is
   done by what signal you send to the server process.
   <variablelist>
    <varlistentry>
2139
     <term><systemitem>SIGTERM</systemitem></term>
2140 2141
     <listitem>
      <para>
2142
       After receiving <systemitem>SIGTERM</systemitem>, the postmaster disallows new
2143 2144 2145
       connections, but lets existing backends end their work normally.
       It shuts down only after all of the backends terminate by client
       request.
2146 2147 2148 2149 2150 2151
       This is the <firstterm>Smart Shutdown</firstterm>.
      </para>
     </listitem>
    </varlistentry>

    <varlistentry>
2152
     <term><systemitem>SIGINT</systemitem></term>
2153 2154
     <listitem>
      <para>
2155
       The postmaster disallows new connections and sends all existing
2156
       backends <systemitem>SIGTERM</systemitem>, which will cause them to abort their current
2157 2158 2159
       transactions and exit promptly.  It then waits for the backends to exit
       and finally shuts down the data base.
       This is the <firstterm>Fast Shutdown</firstterm>.
2160 2161 2162 2163 2164
      </para>
     </listitem>
    </varlistentry>

    <varlistentry>
2165
     <term><systemitem>SIGQUIT</systemitem></term>
2166 2167 2168
     <listitem>
      <para>
       This is the <firstterm>Immediate Shutdown</firstterm> which
2169
       will cause the postmaster to send a <systemitem>SIGQUIT</systemitem> to all backends and
2170
       exit immediately (without properly shutting down the database
2171
       system).  The backends likewise exit immediately upon receiving
2172
       <systemitem>SIGQUIT</systemitem>. This will lead to recovery (by replaying the WAL log)
2173
       upon next start-up.  This is recommended only in emergencies.
2174 2175 2176 2177 2178 2179 2180
      </para>
     </listitem>
    </varlistentry>
   </variablelist>

   <caution>
    <para>
2181
     It is best not to use <systemitem>SIGKILL</systemitem> to shut down the postmaster. This
2182 2183
     will prevent the postmaster from releasing shared memory and
     semaphores, which you may then have to do by hand.
2184 2185 2186
    </para>
   </caution>

2187
   The <acronym>PID</> of the postmaster process can be found using the
2188 2189 2190 2191
   <application>ps</application> program, or from the file
   <filename>postmaster.pid</filename> in the data directory. So for
   example, to do a fast shutdown:
<screen>
2192
&gt; <userinput>kill -INT `head -1 /usr/local/pgsql/data/postmaster.pid`</userinput>
2193 2194 2195 2196
</screen>
  </para>
  <para>
   The program <application>pg_ctl</application> is a shell script
2197 2198
   that provides a more convenient interface for shutting down the
   postmaster.
2199 2200 2201
  </para>
 </sect1>

2202
 <sect1 id="ssl-tcp">
2203
  <title>Secure TCP/IP Connections with SSL</title>
2204

2205 2206 2207 2208
  <indexterm zone="ssl-tcp">
   <primary>SSL</primary>
  </indexterm>

2209
  <para>
2210 2211
   <productname>PostgreSQL</> has native support for connections over
   <acronym>SSL</> to encrypt
2212 2213
   client/server communications for increased security. This requires
   <productname>OpenSSL</productname> to be installed on both client
2214 2215
   and server systems and support enabled at build-time (see <xref
   linkend="installation">).
2216 2217 2218
  </para>

  <para>
2219 2220 2221 2222
   With SSL support compiled in, the <productname>PostgreSQL</> server
   can be started with the argument <option>-l</> (ell) to enable
   SSL connections. When starting in SSL mode, the postmaster will look
   for the files <filename>server.key</> and <filename>server.crt</> in
2223
   the data directory.  These files should contain the server private key
2224 2225 2226
   and certificate respectively. These files must be set up correctly
   before an SSL-enabled server can start. If the private key is protected
   with a passphrase, the postmaster will prompt for the passphrase and will
2227
   not start until it has been entered.
2228 2229 2230 2231 2232
  </para>

  <para>
   The postmaster will listen for both standard and SSL connections
   on the same TCP/IP port, and will negotiate with any connecting
2233
   client whether or not to use SSL.
2234
   See <xref linkend="client-authentication">
2235 2236
   about how to force on the server side the use of SSL for certain
   connections.
2237 2238 2239 2240
  </para>

  <para>
   For details on how to create your server private key and certificate,
2241
   refer to the <productname>OpenSSL</> documentation. A simple self-signed
2242
   certificate can be used to get started for testing, but a certificate signed
2243
   by a <acronym>CA</> (either one of the global <acronym>CAs</> or a local one) should be used in 
2244
   production so the client can verify the servers identity. To create
2245
   a quick self-signed certificate, use the following <productname>OpenSSL</productname> command:
2246 2247 2248
<programlisting>
openssl req -new -text -out cert.req
</programlisting>
2249
   Fill out the information that <command>openssl</> asks for. Make sure that you enter
2250
   the local host name as Common Name; the challenge password can be
2251 2252 2253 2254 2255
   left blank. The script will generate a key that is passphrase protected;
   it will not accept a pass phrase that is less than four characters long.
   To remove the passphrase (as you must if you want automatic start-up of
   the postmaster), run the commands
<programlisting>
2256
openssl rsa -in privkey.pem -out cert.pem
2257
</programlisting>
2258
   Enter the old passphrase to unlock the existing key. Now do
2259 2260 2261 2262 2263
<programlisting>
openssl req -x509 -in cert.req -text -key cert.pem -out cert.cert
cp cert.pem <replaceable>$PGDATA</replaceable>/server.key
cp cert.cert <replaceable>$PGDATA</replaceable>/server.crt
</programlisting>
2264
   to turn the certificate into a self-signed certificate and to copy the
2265
   key and certificate to where the postmaster will look for them.
2266 2267 2268
  </para>
 </sect1>

2269
 <sect1 id="ssh-tunnels">
2270
  <title>Secure TCP/IP Connections with <application>SSH</application> tunnels</title>
2271

2272 2273 2274 2275
  <indexterm zone="ssh-tunnels">
   <primary>ssh</primary>
  </indexterm>

2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
  <note>
   <title>Acknowledgement</title>
   <para>
    Idea taken from an email by Gene Selkov, Jr.
    (<email>selkovjr@mcs.anl.gov</>) written on 1999-09-08 in response
    to a question from Eric Marsden.
   </para>
  </note>

  <para>
   One can use <productname>ssh</productname> to encrypt the network
   connection between clients and a
   <productname>Postgres</productname> server. Done properly, this
   should lead to an adequately secure network connection.
  </para>

  <para>
2293
   First make sure that an <application>ssh</application> server is
2294 2295
   running properly on the same machine as
   <productname>Postgres</productname> and that you can log in using
2296
   <command>ssh</command> as some user. Then you can establish a secure tunnel with a
2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313
   command like this from the client machine:
<programlisting>
&gt; <userinput>ssh -L 3333:foo.com:5432 joe@foo.com</userinput>
</programlisting>
   The first number in the <option>-L</option> argument, 3333, is the
   port number of your end of the tunnel; it can be chosen freely. The
   second number, 5432, is the remote end of the tunnel -- the port
   number your backend is using. The name or the address in between
   the port numbers is the host with the database server you are going
   to connect to. In order to connect to the database server using
   this tunnel, you connect to port 3333 on the local machine:
<programlisting>
psql -h localhost -p 3333 template1
</programlisting>
   To the database server it will then look as though you are really
   user <literal>joe@foo.com</literal> and it will use whatever
   authentication procedure was set up for this user. In order for the
2314 2315
   tunnel setup to succeed you must be allowed to connect via <command>ssh</command> as
   <systemitem>joe@foo.com</systemitem>, just as if you had attempted to use <command>ssh</command> to set up a
2316 2317 2318
   terminal session.
  </para>

2319 2320 2321 2322 2323 2324 2325
  <tip>
   <para>
    Several other products exist that can provide secure tunnels using
    a procedure similar in concept to the one just described.
   </para>
  </tip>

2326
 </sect1>
2327 2328

</Chapter>
2329 2330 2331

<!-- Keep this comment at the end of the file
Local variables:
2332
mode:sgml
2333 2334 2335 2336 2337 2338 2339 2340 2341
sgml-omittag:nil
sgml-shorttag:t
sgml-minimize-attributes:nil
sgml-always-quote-attributes:t
sgml-indent-step:1
sgml-indent-data:t
sgml-parent-document:nil
sgml-default-dtd-file:"./reference.ced"
sgml-exposed-tags:nil
2342
sgml-local-catalogs:("/usr/lib/sgml/catalog")
2343 2344 2345
sgml-local-ecat-files:nil
End:
-->