Baoding-FenShuaJiang
/
Green_GoodERP18_FSJ_Standard
Watch 1
Star 0
Wiki 捐赠
gooderp18绿色标准版
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Tree: 0f75e31513
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '0f75e31513'
${ noResults }
Green_GoodERP18_FSJ_Standard/PostgreSQL/doc/postgresql/html/creating-cluster.html

191 lines
16KB
Raw Blame History 

  1. <?xml version="1.0" encoding="UTF-8" standalone="no"?>

  2. <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>18.2. Creating a Database Cluster</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@lists.postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets V1.79.1" /><link rel="prev" href="postgres-user.html" title="18.1. The PostgreSQL User Account" /><link rel="next" href="server-start.html" title="18.3. Starting the Database Server" /></head><body><div xmlns="http://www.w3.org/TR/xhtml1/transitional" class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">18.2. Creating a Database Cluster</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="postgres-user.html" title="18.1. The PostgreSQL User Account">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="runtime.html" title="Chapter 18. Server Setup and Operation">Up</a></td><th width="60%" align="center">Chapter 18. Server Setup and Operation</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 12.4 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="server-start.html" title="18.3. Starting the Database Server">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="CREATING-CLUSTER"><div class="titlepage"><div><div><h2 class="title" style="clear: both">18.2. Creating a Database Cluster</h2></div></div></div><div class="toc"><dl class="toc"><dt><span class="sect2"><a href="creating-cluster.html#CREATING-CLUSTER-MOUNT-POINTS">18.2.1. Use of Secondary File Systems</a></span></dt><dt><span class="sect2"><a href="creating-cluster.html#CREATING-CLUSTER-FILESYSTEM">18.2.2. File Systems</a></span></dt></dl></div><a id="id-1.6.5.4.2" class="indexterm"></a><a id="id-1.6.5.4.3" class="indexterm"></a><p>

  3.    Before you can do anything, you must initialize a database storage

  4.    area on disk. We call this a <em class="firstterm">database cluster</em>.

  5.    (The <acronym class="acronym">SQL</acronym> standard uses the term catalog cluster.) A

  6.    database cluster is a collection of databases that is managed by a

  7.    single instance of a running database server. After initialization, a

  8.    database cluster will contain a database named <code class="literal">postgres</code>,

  9.    which is meant as a default database for use by utilities, users and third

  10.    party applications.  The database server itself does not require the

  11.    <code class="literal">postgres</code> database to exist, but many external utility

  12.    programs assume it exists.  Another database created within each cluster

  13.    during initialization is called

  14.    <code class="literal">template1</code>.  As the name suggests, this will be used

  15.    as a template for subsequently created databases; it should not be

  16.    used for actual work.  (See <a class="xref" href="managing-databases.html" title="Chapter 22. Managing Databases">Chapter 22</a> for

  17.    information about creating new databases within a cluster.)

  18.   </p><p>

  19.    In file system terms, a database cluster is a single directory

  20.    under which all data will be stored. We call this the <em class="firstterm">data

  21.    directory</em> or <em class="firstterm">data area</em>. It is

  22.    completely up to you where you choose to store your data.  There is no

  23.    default, although locations such as

  24.    <code class="filename">/usr/local/pgsql/data</code> or

  25.    <code class="filename">/var/lib/pgsql/data</code> are popular. To initialize a

  26.    database cluster, use the command <a class="xref" href="app-initdb.html" title="initdb"><span class="refentrytitle">initdb</span></a>,<a id="id-1.6.5.4.5.6" class="indexterm"></a> which is

  27.    installed with <span class="productname">PostgreSQL</span>. The desired

  28.    file system location of your database cluster is indicated by the

  29.    <code class="option">-D</code> option, for example:

  30. </p><pre class="screen">

  31. <code class="prompt">$</code> <strong class="userinput"><code>initdb -D /usr/local/pgsql/data</code></strong>

  32. </pre><p>

  33.    Note that you must execute this command while logged into the

  34.    <span class="productname">PostgreSQL</span> user account, which is

  35.    described in the previous section.

  36.   </p><div class="tip"><h3 class="title">Tip</h3><p>

  37.     As an alternative to the <code class="option">-D</code> option, you can set

  38.     the environment variable <code class="envar">PGDATA</code>.

  39.     <a id="id-1.6.5.4.6.1.3" class="indexterm"></a>

  40.    </p></div><p>

  41.    Alternatively, you can run <code class="command">initdb</code> via

  42.    the <a class="xref" href="app-pg-ctl.html" title="pg_ctl"><span class="refentrytitle"><span class="application">pg_ctl</span></span></a>

  43.    program<a id="id-1.6.5.4.7.3" class="indexterm"></a> like so:

  44. </p><pre class="screen">

  45. <code class="prompt">$</code> <strong class="userinput"><code>pg_ctl -D /usr/local/pgsql/data initdb</code></strong>

  46. </pre><p>

  47.    This may be more intuitive if you are

  48.    using <code class="command">pg_ctl</code> for starting and stopping the

  49.    server (see <a class="xref" href="server-start.html" title="18.3. Starting the Database Server">Section 18.3</a>), so

  50.    that <code class="command">pg_ctl</code> would be the sole command you use

  51.    for managing the database server instance.

  52.   </p><p>

  53.    <code class="command">initdb</code> will attempt to create the directory you

  54.    specify if it does not already exist.  Of course, this will fail if

  55.    <code class="command">initdb</code> does not have permissions to write in the

  56.    parent directory.  It's generally recommendable that the

  57.    <span class="productname">PostgreSQL</span> user own not just the data

  58.    directory but its parent directory as well, so that this should not

  59.    be a problem.  If the desired parent directory doesn't exist either,

  60.    you will need to create it first, using root privileges if the

  61.    grandparent directory isn't writable.  So the process might look

  62.    like this:

  63. </p><pre class="screen">

  64. root# <strong class="userinput"><code>mkdir /usr/local/pgsql</code></strong>

  65. root# <strong class="userinput"><code>chown postgres /usr/local/pgsql</code></strong>

  66. root# <strong class="userinput"><code>su postgres</code></strong>

  67. postgres$ <strong class="userinput"><code>initdb -D /usr/local/pgsql/data</code></strong>

  68. </pre><p>

  69.   </p><p>

  70.    <code class="command">initdb</code> will refuse to run if the data directory

  71.    exists and already contains files; this is to prevent accidentally

  72.    overwriting an existing installation.

  73.   </p><p>

  74.    Because the data directory contains all the data stored in the

  75.    database, it is essential that it be secured from unauthorized

  76.    access. <code class="command">initdb</code> therefore revokes access

  77.    permissions from everyone but the

  78.    <span class="productname">PostgreSQL</span> user, and optionally, group.

  79.    Group access, when enabled, is read-only.  This allows an unprivileged

  80.    user in the same group as the cluster owner to take a backup of the

  81.    cluster data or perform other operations that only require read access.

  82.   </p><p>

  83.    Note that enabling or disabling group access on an existing cluster requires

  84.    the cluster to be shut down and the appropriate mode to be set on all

  85.    directories and files before restarting

  86.    <span class="productname">PostgreSQL</span>.  Otherwise, a mix of modes might

  87.    exist in the data directory.  For clusters that allow access only by the

  88.    owner, the appropriate modes are <code class="literal">0700</code> for directories

  89.    and <code class="literal">0600</code> for files.  For clusters that also allow

  90.    reads by the group, the appropriate modes are <code class="literal">0750</code>

  91.    for directories and <code class="literal">0640</code> for files.

  92.   </p><p>

  93.    However, while the directory contents are secure, the default

  94.    client authentication setup allows any local user to connect to the

  95.    database and even become the database superuser. If you do not

  96.    trust other local users, we recommend you use one of

  97.    <code class="command">initdb</code>'s <code class="option">-W</code>, <code class="option">--pwprompt</code>

  98.    or <code class="option">--pwfile</code> options to assign a password to the

  99.    database superuser.<a id="id-1.6.5.4.12.5" class="indexterm"></a>

  100.    Also, specify <code class="option">-A md5</code> or

  101.    <code class="option">-A password</code> so that the default <code class="literal">trust</code> authentication

  102.    mode is not used; or modify the generated <code class="filename">pg_hba.conf</code>

  103.    file after running <code class="command">initdb</code>, but

  104.    <span class="emphasis"><em>before</em></span> you start the server for the first time. (Other

  105.    reasonable approaches include using <code class="literal">peer</code> authentication

  106.    or file system permissions to restrict connections. See <a class="xref" href="client-authentication.html" title="Chapter 20. Client Authentication">Chapter 20</a> for more information.)

  107.   </p><p>

  108.    <code class="command">initdb</code> also initializes the default

  109.    locale<a id="id-1.6.5.4.13.2" class="indexterm"></a> for the database cluster.

  110.    Normally, it will just take the locale settings in the environment

  111.    and apply them to the initialized database.  It is possible to

  112.    specify a different locale for the database; more information about

  113.    that can be found in <a class="xref" href="locale.html" title="23.1. Locale Support">Section 23.1</a>.  The default sort order used

  114.    within the particular database cluster is set by

  115.    <code class="command">initdb</code>, and while you can create new databases using

  116.    different sort order, the order used in the template databases that initdb

  117.    creates cannot be changed without dropping and recreating them.

  118.    There is also a performance impact for using locales

  119.    other than <code class="literal">C</code> or <code class="literal">POSIX</code>. Therefore, it is

  120.    important to make this choice correctly the first time.

  121.   </p><p>

  122.    <code class="command">initdb</code> also sets the default character set encoding

  123.    for the database cluster.  Normally this should be chosen to match the

  124.    locale setting.  For details see <a class="xref" href="multibyte.html" title="23.3. Character Set Support">Section 23.3</a>.

  125.   </p><p>

  126.    Non-<code class="literal">C</code> and non-<code class="literal">POSIX</code> locales rely on the

  127.    operating system's collation library for character set ordering.

  128.    This controls the ordering of keys stored in indexes.  For this reason,

  129.    a cluster cannot switch to an incompatible collation library version,

  130.    either through snapshot restore, binary streaming replication, a

  131.    different operating system, or an operating system upgrade.

  132.   </p><div class="sect2" id="CREATING-CLUSTER-MOUNT-POINTS"><div class="titlepage"><div><div><h3 class="title">18.2.1. Use of Secondary File Systems</h3></div></div></div><a id="id-1.6.5.4.16.2" class="indexterm"></a><p>

  133.     Many installations create their database clusters on file systems

  134.     (volumes) other than the machine's <span class="quote">“<span class="quote">root</span>”</span> volume.  If you

  135.     choose to do this, it is not advisable to try to use the secondary

  136.     volume's topmost directory (mount point) as the data directory.

  137.     Best practice is to create a directory within the mount-point

  138.     directory that is owned by the <span class="productname">PostgreSQL</span>

  139.     user, and then create the data directory within that.  This avoids

  140.     permissions problems, particularly for operations such

  141.     as <span class="application">pg_upgrade</span>, and it also ensures clean failures if

  142.     the secondary volume is taken offline.

  143.    </p></div><div class="sect2" id="CREATING-CLUSTER-FILESYSTEM"><div class="titlepage"><div><div><h3 class="title">18.2.2. File Systems</h3></div></div></div><p>

  144.     Generally, any file system with POSIX semantics can be used for

  145.     PostgreSQL.  Users prefer different file systems for a variety of reasons,

  146.     including vendor support, performance, and familiarity.  Experience

  147.     suggests that, all other things being equal, one should not expect major

  148.     performance or behavior changes merely from switching file systems or

  149.     making minor file system configuration changes.

  150.    </p><div class="sect3" id="CREATING-CLUSTER-NFS"><div class="titlepage"><div><div><h4 class="title">18.2.2.1. NFS</h4></div></div></div><a id="id-1.6.5.4.17.3.2" class="indexterm"></a><p>

  151.      It is possible to use an <acronym class="acronym">NFS</acronym> file system for storing

  152.      the <span class="productname">PostgreSQL</span> data directory.

  153.      <span class="productname">PostgreSQL</span> does nothing special for

  154.      <acronym class="acronym">NFS</acronym> file systems, meaning it assumes

  155.      <acronym class="acronym">NFS</acronym> behaves exactly like locally-connected drives.

  156.      <span class="productname">PostgreSQL</span> does not use any functionality that

  157.      is known to have nonstandard behavior on <acronym class="acronym">NFS</acronym>, such as

  158.      file locking.

  159.     </p><p>

  160.      The only firm requirement for using <acronym class="acronym">NFS</acronym> with

  161.      <span class="productname">PostgreSQL</span> is that the file system is mounted

  162.      using the <code class="literal">hard</code> option.  With the

  163.      <code class="literal">hard</code> option, processes can <span class="quote">“<span class="quote">hang</span>”</span>

  164.      indefinitely if there are network problems, so this configuration will

  165.      require a careful monitoring setup.  The <code class="literal">soft</code> option

  166.      will interrupt system calls in case of network problems, but

  167.      <span class="productname">PostgreSQL</span> will not repeat system calls

  168.      interrupted in this way, so any such interruption will result in an I/O

  169.      error being reported.

  170.     </p><p>

  171.      It is not necessary to use the <code class="literal">sync</code> mount option.  The

  172.      behavior of the <code class="literal">async</code> option is sufficient, since

  173.      <span class="productname">PostgreSQL</span> issues <code class="literal">fsync</code>

  174.      calls at appropriate times to flush the write caches.  (This is analogous

  175.      to how it works on a local file system.)  However, it is strongly

  176.      recommended to use the <code class="literal">sync</code> export option on the NFS

  177.      <span class="emphasis"><em>server</em></span> on systems where it exists (mainly Linux).

  178.      Otherwise, an <code class="literal">fsync</code> or equivalent on the NFS client is

  179.      not actually guaranteed to reach permanent storage on the server, which

  180.      could cause corruption similar to running with the parameter <a class="xref" href="runtime-config-wal.html#GUC-FSYNC">fsync</a> off.  The defaults of these mount and export

  181.      options differ between vendors and versions, so it is recommended to

  182.      check and perhaps specify them explicitly in any case to avoid any

  183.      ambiguity.

  184.     </p><p>

  185.      In some cases, an external storage product can be accessed either via NFS

  186.      or a lower-level protocol such as iSCSI.  In the latter case, the storage

  187.      appears as a block device and any available file system can be created on

  188.      it.  That approach might relieve the DBA from having to deal with some of

  189.      the idiosyncrasies of NFS, but of course the complexity of managing

  190.      remote storage then happens at other levels.

  191.     </p></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="postgres-user.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="runtime.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="server-start.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">18.1. The <span class="productname">PostgreSQL</span> User Account </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> 18.3. Starting the Database Server</td></tr></table></div></body></html>
上海开阖软件有限公司 沪ICP备12045867号-1