Baoding-FenShuaJiang
/
Green_GoodERP18_FSJ_Standard
關註 1
收藏 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.
分支: master
分支列表 標籤列表
${ item.name }
Create branch ${ searchTerm }
from 'master'
${ noResults }
Green_GoodERP18_FSJ_Standard/PostgreSQL/doc/postgresql/html/pgupgrade.html

424 line
35KB
原始文件 永久連結 Blame 文件歷史 

  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>pg_upgrade</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="pgtesttiming.html" title="pg_test_timing" /><link rel="next" href="pgwaldump.html" title="pg_waldump" /></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"><span xmlns="http://www.w3.org/1999/xhtml" class="application">pg_upgrade</span></th></tr><tr><td width="10%" align="left"><a accesskey="p" href="pgtesttiming.html" title="pg_test_timing">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="reference-server.html" title="PostgreSQL Server Applications">Up</a></td><th width="60%" align="center">PostgreSQL Server Applications</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="pgwaldump.html" title="pg_waldump">Next</a></td></tr></table><hr></hr></div><div class="refentry" id="PGUPGRADE"><div class="titlepage"></div><a id="id-1.9.5.12.1" class="indexterm"></a><div class="refnamediv"><h2><span class="refentrytitle"><span class="application">pg_upgrade</span></span></h2><p>pg_upgrade — upgrade a <span class="productname">PostgreSQL</span> server instance</p></div><div class="refsynopsisdiv"><h2>Synopsis</h2><div class="cmdsynopsis"><p id="id-1.9.5.12.4.1"><code class="command">pg_upgrade</code>  <code class="option">-b</code>   <em class="replaceable"><code>oldbindir</code></em>   <code class="option">-B</code>   <em class="replaceable"><code>newbindir</code></em>   <code class="option">-d</code>   <em class="replaceable"><code>oldconfigdir</code></em>   <code class="option">-D</code>   <em class="replaceable"><code>newconfigdir</code></em>  [<em class="replaceable"><code>option</code></em>...]</p></div></div><div class="refsect1" id="id-1.9.5.12.5"><h2>Description</h2><p>

  3.   <span class="application">pg_upgrade</span> (formerly called <span class="application">pg_migrator</span>) allows data

  4.   stored in <span class="productname">PostgreSQL</span> data files to be upgraded to a later <span class="productname">PostgreSQL</span>

  5.   major version without the data dump/reload typically required for

  6.   major version upgrades, e.g. from 9.5.8 to 9.6.4 or from 10.7 to 11.2.

  7.   It is not required for minor version upgrades, e.g. from 9.6.2 to 9.6.3

  8.   or from 10.1 to 10.2.

  9.  </p><p>

  10.   Major PostgreSQL releases regularly add new features that often

  11.   change the layout of the system tables, but the internal data storage

  12.   format rarely changes.  <span class="application">pg_upgrade</span> uses this fact

  13.   to perform rapid upgrades by creating new system tables and simply

  14.   reusing the old user data files.  If a future major release ever

  15.   changes the data storage format in a way that makes the old data

  16.   format unreadable, <span class="application">pg_upgrade</span> will not be usable

  17.   for such upgrades.  (The community will attempt to avoid such

  18.   situations.)

  19.  </p><p>

  20.   <span class="application">pg_upgrade</span> does its best to

  21.   make sure the old and new clusters are binary-compatible, e.g.  by

  22.   checking for compatible compile-time settings, including 32/64-bit

  23.   binaries.  It is important that

  24.   any external modules are also binary compatible, though this cannot

  25.   be checked by <span class="application">pg_upgrade</span>.

  26.  </p><p>

  27.    pg_upgrade supports upgrades from 8.4.X and later to the current

  28.    major release of <span class="productname">PostgreSQL</span>, including snapshot and beta releases.

  29.   </p></div><div class="refsect1" id="id-1.9.5.12.6"><h2>Options</h2><p>

  30.     <span class="application">pg_upgrade</span> accepts the following command-line arguments:

  31. 

  32.     </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="option">-b</code> <em class="replaceable"><code>bindir</code></em><br /></span><span class="term"><code class="option">--old-bindir=</code><em class="replaceable"><code>bindir</code></em></span></dt><dd><p>the old PostgreSQL executable directory;

  33.       environment variable <code class="envar">PGBINOLD</code></p></dd><dt><span class="term"><code class="option">-B</code> <em class="replaceable"><code>bindir</code></em><br /></span><span class="term"><code class="option">--new-bindir=</code><em class="replaceable"><code>bindir</code></em></span></dt><dd><p>the new PostgreSQL executable directory;

  34.       environment variable <code class="envar">PGBINNEW</code></p></dd><dt><span class="term"><code class="option">-c</code><br /></span><span class="term"><code class="option">--check</code></span></dt><dd><p>check clusters only, don't change any data</p></dd><dt><span class="term"><code class="option">-d</code> <em class="replaceable"><code>configdir</code></em><br /></span><span class="term"><code class="option">--old-datadir=</code><em class="replaceable"><code>configdir</code></em></span></dt><dd><p>the old database cluster configuration directory; environment

  35.       variable <code class="envar">PGDATAOLD</code></p></dd><dt><span class="term"><code class="option">-D</code> <em class="replaceable"><code>configdir</code></em><br /></span><span class="term"><code class="option">--new-datadir=</code><em class="replaceable"><code>configdir</code></em></span></dt><dd><p>the new database cluster configuration directory; environment

  36.       variable <code class="envar">PGDATANEW</code></p></dd><dt><span class="term"><code class="option">-j <em class="replaceable"><code>njobs</code></em></code><br /></span><span class="term"><code class="option">--jobs=<em class="replaceable"><code>njobs</code></em></code></span></dt><dd><p>number of simultaneous processes or threads to use

  37.       </p></dd><dt><span class="term"><code class="option">-k</code><br /></span><span class="term"><code class="option">--link</code></span></dt><dd><p>use hard links instead of copying files to the new

  38.       cluster</p></dd><dt><span class="term"><code class="option">-o</code> <em class="replaceable"><code>options</code></em><br /></span><span class="term"><code class="option">--old-options</code> <em class="replaceable"><code>options</code></em></span></dt><dd><p>options to be passed directly to the

  39.       old <code class="command">postgres</code> command;  multiple

  40.       option invocations are appended</p></dd><dt><span class="term"><code class="option">-O</code> <em class="replaceable"><code>options</code></em><br /></span><span class="term"><code class="option">--new-options</code> <em class="replaceable"><code>options</code></em></span></dt><dd><p>options to be passed directly to the

  41.       new <code class="command">postgres</code> command;  multiple

  42.       option invocations are appended</p></dd><dt><span class="term"><code class="option">-p</code> <em class="replaceable"><code>port</code></em><br /></span><span class="term"><code class="option">--old-port=</code><em class="replaceable"><code>port</code></em></span></dt><dd><p>the old cluster port number; environment

  43.       variable <code class="envar">PGPORTOLD</code></p></dd><dt><span class="term"><code class="option">-P</code> <em class="replaceable"><code>port</code></em><br /></span><span class="term"><code class="option">--new-port=</code><em class="replaceable"><code>port</code></em></span></dt><dd><p>the new cluster port number; environment

  44.       variable <code class="envar">PGPORTNEW</code></p></dd><dt><span class="term"><code class="option">-r</code><br /></span><span class="term"><code class="option">--retain</code></span></dt><dd><p>retain SQL and log files even after successful completion

  45.       </p></dd><dt><span class="term"><code class="option">-s</code> <em class="replaceable"><code>dir</code></em><br /></span><span class="term"><code class="option">--socketdir=</code><em class="replaceable"><code>dir</code></em></span></dt><dd><p>directory to use for postmaster sockets during upgrade;

  46.       default is current working directory; environment

  47.       variable <code class="envar">PGSOCKETDIR</code></p></dd><dt><span class="term"><code class="option">-U</code> <em class="replaceable"><code>username</code></em><br /></span><span class="term"><code class="option">--username=</code><em class="replaceable"><code>username</code></em></span></dt><dd><p>cluster's install user name; environment

  48.       variable <code class="envar">PGUSER</code></p></dd><dt><span class="term"><code class="option">-v</code><br /></span><span class="term"><code class="option">--verbose</code></span></dt><dd><p>enable verbose internal logging</p></dd><dt><span class="term"><code class="option">-V</code><br /></span><span class="term"><code class="option">--version</code></span></dt><dd><p>display version information, then exit</p></dd><dt><span class="term"><code class="option">--clone</code></span></dt><dd><p>

  49.         Use efficient file cloning (also known as <span class="quote">“<span class="quote">reflinks</span>”</span> on

  50.         some systems) instead of copying files to the new cluster.  This can

  51.         result in near-instantaneous copying of the data files, giving the

  52.         speed advantages of <code class="option">-k</code>/<code class="option">--link</code> while

  53.         leaving the old cluster untouched.

  54.        </p><p>

  55.         File cloning is only supported on some operating systems and file

  56.         systems.  If it is selected but not supported, the

  57.         <span class="application">pg_upgrade</span> run will error.  At present, it

  58.         is supported on Linux (kernel 4.5 or later) with Btrfs and XFS (on

  59.         file systems created with reflink support), and on macOS with APFS.

  60.        </p></dd><dt><span class="term"><code class="option">-?</code><br /></span><span class="term"><code class="option">--help</code></span></dt><dd><p>show help, then exit</p></dd></dl></div><p>

  61.    </p></div><div class="refsect1" id="id-1.9.5.12.7"><h2>Usage</h2><p>

  62.    These are the steps to perform an upgrade

  63.    with <span class="application">pg_upgrade</span>:

  64.   </p><div class="procedure"><ol class="procedure" type="1"><li class="step"><p class="title"><strong>Optionally move the old cluster</strong></p><p>

  65.      If you are using a version-specific installation directory, e.g.

  66.      <code class="filename">/opt/PostgreSQL/12</code>, you do not need to move the old cluster. The

  67.      graphical installers all use version-specific installation directories.

  68.     </p><p>

  69.      If your installation directory is not version-specific, e.g.

  70.      <code class="filename">/usr/local/pgsql</code>, it is necessary to move the current PostgreSQL install

  71.      directory so it does not interfere with the new <span class="productname">PostgreSQL</span> installation.

  72.      Once the current <span class="productname">PostgreSQL</span> server is shut down, it is safe to rename the

  73.      PostgreSQL installation directory; assuming the old directory is

  74.      <code class="filename">/usr/local/pgsql</code>, you can do:

  75. 

  76. </p><pre class="programlisting">

  77. mv /usr/local/pgsql /usr/local/pgsql.old

  78. </pre><p>

  79.      to rename the directory.

  80.     </p></li><li class="step"><p class="title"><strong>For source installs, build the new version</strong></p><p>

  81.      Build the new PostgreSQL source with <code class="command">configure</code> flags that are compatible

  82.      with the old cluster. <span class="application">pg_upgrade</span> will check <code class="command">pg_controldata</code> to make

  83.      sure all settings are compatible before starting the upgrade.

  84.     </p></li><li class="step"><p class="title"><strong>Install the new PostgreSQL binaries</strong></p><p>

  85.      Install the new server's binaries and support

  86.      files.  <span class="application">pg_upgrade</span> is included in a default installation.

  87.     </p><p>

  88.      For source installs, if you wish to install the new server in a custom

  89.      location, use the <code class="literal">prefix</code> variable:

  90. 

  91. </p><pre class="programlisting">

  92. make prefix=/usr/local/pgsql.new install

  93. </pre></li><li class="step"><p class="title"><strong>Initialize the new PostgreSQL cluster</strong></p><p>

  94.      Initialize the new cluster using <code class="command">initdb</code>.

  95.      Again, use compatible <code class="command">initdb</code>

  96.      flags that match the old cluster. Many

  97.      prebuilt installers do this step automatically. There is no need to

  98.      start the new cluster.

  99.     </p></li><li class="step"><p class="title"><strong>Install custom shared object files</strong></p><p>

  100.      Install any custom shared object files (or DLLs) used by the old cluster

  101.      into the new cluster, e.g. <code class="filename">pgcrypto.so</code>,

  102.      whether they are from <code class="filename">contrib</code>

  103.      or some other source. Do not install the schema definitions, e.g.

  104.      <code class="command">CREATE EXTENSION pgcrypto</code>, because these will be upgraded

  105.      from the old cluster.

  106.      Also, any custom full text search files (dictionary, synonym,

  107.      thesaurus, stop words) must also be copied to the new cluster.

  108.     </p></li><li class="step"><p class="title"><strong>Adjust authentication</strong></p><p>

  109.      <code class="command">pg_upgrade</code> will connect to the old and new servers several

  110.      times, so you might want to set authentication to <code class="literal">peer</code>

  111.      in <code class="filename">pg_hba.conf</code> or use a <code class="filename">~/.pgpass</code> file

  112.      (see <a class="xref" href="libpq-pgpass.html" title="33.15. The Password File">Section 33.15</a>).

  113.     </p></li><li class="step"><p class="title"><strong>Stop both servers</strong></p><p>

  114.      Make sure both database servers are stopped using, on Unix, e.g.:

  115. 

  116. </p><pre class="programlisting">

  117. pg_ctl -D /opt/PostgreSQL/9.6 stop

  118. pg_ctl -D /opt/PostgreSQL/12 stop

  119. </pre><p>

  120. 

  121.      or on Windows, using the proper service names:

  122. 

  123. </p><pre class="programlisting">

  124. NET STOP postgresql-9.6

  125. NET STOP postgresql-12

  126. </pre><p>

  127.     </p><p>

  128.      Streaming replication and log-shipping standby servers can

  129.      remain running until a later step.

  130.     </p></li><li class="step"><p class="title"><strong>Prepare for standby server upgrades</strong></p><p>

  131.      If you are upgrading standby servers using methods outlined in section <a class="xref" href="pgupgrade.html#PGUPGRADE-STEP-REPLICAS" title="Upgrade streaming replication and log-shipping standby servers">Step 10</a>, verify that the old standby

  132.      servers are caught up by running <span class="application">pg_controldata</span>

  133.      against the old primary and standby clusters.  Verify that the

  134.      <span class="quote">“<span class="quote">Latest checkpoint location</span>”</span> values match in all clusters.

  135.      (There will be a mismatch if old standby servers were shut down

  136.      before the old primary or if the old standby servers are still running.)

  137.      Also, make sure <code class="varname">wal_level</code> is not set to 

  138.      <code class="literal">minimal</code> in the <code class="filename">postgresql.conf</code> file on the

  139.      new primary cluster.

  140.     </p></li><li class="step"><p class="title"><strong>Run <span class="application">pg_upgrade</span></strong></p><p>

  141.      Always run the <span class="application">pg_upgrade</span> binary of the new server, not the old one.

  142.      <span class="application">pg_upgrade</span> requires the specification of the old and new cluster's

  143.      data and executable (<code class="filename">bin</code>) directories. You can also specify

  144.      user and port values, and whether you want the data files linked or cloned

  145.      instead of the default copy behavior.

  146.     </p><p>

  147.      If you use link mode, the upgrade will be much faster (no file

  148.      copying) and use less disk space, but you will not be able to access

  149.      your old cluster

  150.      once you start the new cluster after the upgrade.  Link mode also

  151.      requires that the old and new cluster data directories be in the

  152.      same file system.  (Tablespaces and <code class="filename">pg_wal</code> can be on

  153.      different file systems.)

  154.      Clone mode provides the same speed and disk space advantages but

  155.      does not cause the old cluster to be unusable once the new cluster

  156.      is started.  Clone mode also requires that the old and new data

  157.      directories be in the same file system.  This mode is only available

  158.      on certain operating systems and file systems.

  159.     </p><p>

  160.      The <code class="option">--jobs</code> option allows multiple CPU cores to be used

  161.      for copying/linking of files and to dump and reload database schemas

  162.      in parallel;  a good place to start is the maximum of the number of

  163.      CPU cores and tablespaces.  This option can dramatically reduce the

  164.      time to upgrade a multi-database server running on a multiprocessor

  165.      machine.

  166.     </p><p>

  167.      For Windows users, you must be logged into an administrative account, and

  168.      then start a shell as the <code class="literal">postgres</code> user and set the proper path:

  169. 

  170. </p><pre class="programlisting">

  171. RUNAS /USER:postgres "CMD.EXE"

  172. SET PATH=%PATH%;C:\Program Files\PostgreSQL\12\bin;

  173. </pre><p>

  174. 

  175.      and then run <span class="application">pg_upgrade</span> with quoted directories, e.g.:

  176. 

  177. </p><pre class="programlisting">

  178. pg_upgrade.exe

  179.         --old-datadir "C:/Program Files/PostgreSQL/9.6/data"

  180.         --new-datadir "C:/Program Files/PostgreSQL/12/data"

  181.         --old-bindir "C:/Program Files/PostgreSQL/9.6/bin"

  182.         --new-bindir "C:/Program Files/PostgreSQL/12/bin"

  183. </pre><p>

  184. 

  185.      Once started, <code class="command">pg_upgrade</code> will verify the two clusters are compatible

  186.      and then do the upgrade. You can use <code class="command">pg_upgrade --check</code>

  187.      to perform only the checks, even if the old server is still

  188.      running. <code class="command">pg_upgrade --check</code> will also outline any

  189.      manual adjustments you will need to make after the upgrade.  If you

  190.      are going to be using link or clone mode, you should use the option

  191.      <code class="option">--link</code> or <code class="option">--clone</code> with

  192.      <code class="option">--check</code> to enable mode-specific checks.

  193.      <code class="command">pg_upgrade</code> requires write permission in the current directory.

  194.     </p><p>

  195.      Obviously, no one should be accessing the clusters during the

  196.      upgrade.  <span class="application">pg_upgrade</span> defaults to running servers

  197.      on port 50432 to avoid unintended client connections.

  198.      You can use the same port number for both clusters when doing an

  199.      upgrade because the old and new clusters will not be running at the

  200.      same time.  However, when checking an old running server, the old

  201.      and new port numbers must be different.

  202.     </p><p>

  203.      If an error occurs while restoring the database schema, <code class="command">pg_upgrade</code> will

  204.      exit and you will have to revert to the old cluster as outlined in <a class="xref" href="pgupgrade.html#PGUPGRADE-STEP-REVERT" title="Reverting to old cluster">Step 16</a>

  205.      below. To try <code class="command">pg_upgrade</code> again, you will need to modify the old

  206.      cluster so the pg_upgrade schema restore succeeds. If the problem is a

  207.      <code class="filename">contrib</code> module, you might need to uninstall the <code class="filename">contrib</code> module from

  208.      the old cluster and install it in the new cluster after the upgrade,

  209.      assuming the module is not being used to store user data.

  210.     </p></li><li class="step" id="PGUPGRADE-STEP-REPLICAS"><p class="title"><strong>Upgrade streaming replication and log-shipping standby servers</strong></p><p>

  211.      If you used link mode and have Streaming Replication (see <a class="xref" href="warm-standby.html#STREAMING-REPLICATION" title="26.2.5. Streaming Replication">Section 26.2.5</a>) or Log-Shipping (see <a class="xref" href="warm-standby.html" title="26.2. Log-Shipping Standby Servers">Section 26.2</a>) standby servers, you can follow these steps to

  212.      quickly upgrade them.  You will not be running <span class="application">pg_upgrade</span> on

  213.      the standby servers, but rather <span class="application">rsync</span> on the primary.

  214.      Do not start any servers yet.

  215.     </p><p>

  216.      If you did <span class="emphasis"><em>not</em></span> use link mode, do not have or do not

  217.      want to use <span class="application">rsync</span>, or want an easier solution, skip

  218.      the instructions in this section and simply recreate the standby

  219.      servers once <span class="application">pg_upgrade</span> completes and the new primary

  220.      is running.

  221.     </p><ol type="a" class="substeps"><li class="step"><p class="title"><strong>Install the new PostgreSQL binaries on standby servers</strong></p><p>

  222.        Make sure the new binaries and support files are installed on all

  223.        standby servers.

  224.       </p></li><li class="step"><p class="title"><strong>Make sure the new standby data directories do <span class="emphasis"><em>not</em></span> exist</strong></p><p>

  225.        Make sure the new standby data directories do <span class="emphasis"><em>not</em></span>

  226.        exist or are empty.  If <span class="application">initdb</span> was run, delete

  227.        the standby servers' new data directories.

  228.       </p></li><li class="step"><p class="title"><strong>Install custom shared object files</strong></p><p>

  229.        Install the same custom shared object files on the new standbys

  230.        that you installed in the new primary cluster.

  231.       </p></li><li class="step"><p class="title"><strong>Stop standby servers</strong></p><p>

  232.        If the standby servers are still running, stop them now using the

  233.        above instructions.

  234.       </p></li><li class="step"><p class="title"><strong>Save configuration files</strong></p><p>

  235.        Save any configuration files from the old standbys' configuration

  236.        directories you need to keep, e.g.  <code class="filename">postgresql.conf</code>,

  237.        <code class="literal">pg_hba.conf</code>, because these will be overwritten or

  238.        removed in the next step.

  239.       </p></li><li class="step"><p class="title"><strong>Run <span class="application">rsync</span></strong></p><p>

  240.        When using link mode, standby servers can be quickly upgraded using

  241.        <span class="application">rsync</span>.  To accomplish this, from a directory on

  242.        the primary server that is above the old and new database cluster

  243.        directories, run this on the <span class="emphasis"><em>primary</em></span> for each standby

  244.        server:

  245. 

  246. </p><pre class="programlisting">

  247. rsync --archive --delete --hard-links --size-only --no-inc-recursive old_cluster new_cluster remote_dir

  248. </pre><p>

  249. 

  250.        where <code class="option">old_cluster</code> and <code class="option">new_cluster</code> are relative

  251.        to the current directory on the primary, and <code class="option">remote_dir</code>

  252.        is <span class="emphasis"><em>above</em></span> the old and new cluster directories

  253.        on the standby.  The directory structure under the specified

  254.        directories on the primary and standbys must match.  Consult the

  255.        <span class="application">rsync</span> manual page for details on specifying the

  256.        remote directory, e.g.

  257. 

  258. </p><pre class="programlisting">

  259. rsync --archive --delete --hard-links --size-only --no-inc-recursive /opt/PostgreSQL/9.5 \

  260.       /opt/PostgreSQL/9.6 standby.example.com:/opt/PostgreSQL

  261. </pre><p>

  262. 

  263.        You can verify what the command will do using

  264.        <span class="application">rsync</span>'s <code class="option">--dry-run</code> option.  While

  265.        <span class="application">rsync</span> must be run on the primary for at least one

  266.        standby, it is possible to run <span class="application">rsync</span> on an upgraded

  267.        standby to upgrade other standbys, as long as the upgraded standby

  268.        has not been started.

  269.       </p><p>

  270.        What this does is to record the links created by

  271.        <span class="application">pg_upgrade</span>'s link mode that connect files in the

  272.        old and new clusters on the primary server.  It then finds matching

  273.        files in the standby's old cluster and creates links for them in the

  274.        standby's new cluster.  Files that were not linked on the primary

  275.        are copied from the primary to the standby.  (They are usually

  276.        small.)  This provides rapid standby upgrades.  Unfortunately,

  277.        <span class="application">rsync</span> needlessly copies files associated with

  278.        temporary and unlogged tables because these files don't normally

  279.        exist on standby servers.

  280.       </p><p>

  281.        If you have tablespaces, you will need to run a similar

  282.        <span class="application">rsync</span> command for each tablespace directory, e.g.:

  283. 

  284. </p><pre class="programlisting">

  285. rsync --archive --delete --hard-links --size-only --no-inc-recursive /vol1/pg_tblsp/PG_9.5_201510051 \

  286.       /vol1/pg_tblsp/PG_9.6_201608131 standby.example.com:/vol1/pg_tblsp

  287. </pre><p>

  288. 

  289.        If you have relocated <code class="filename">pg_wal</code> outside the data

  290.        directories, <span class="application">rsync</span> must be run on those directories

  291.        too.

  292.       </p></li><li class="step"><p class="title"><strong>Configure streaming replication and log-shipping standby servers</strong></p><p>

  293.        Configure the servers for log shipping.  (You do not need to run

  294.        <code class="function">pg_start_backup()</code> and <code class="function">pg_stop_backup()</code>

  295.        or take a file system backup as the standbys are still synchronized

  296.        with the primary.)

  297.       </p></li></ol></li><li class="step"><p class="title"><strong>Restore <code class="filename">pg_hba.conf</code></strong></p><p>

  298.      If you modified <code class="filename">pg_hba.conf</code>, restore its original settings.

  299.      It might also be necessary to adjust other configuration files in the new

  300.      cluster to match the old cluster, e.g. <code class="filename">postgresql.conf</code>.

  301.     </p></li><li class="step"><p class="title"><strong>Start the new server</strong></p><p>

  302.      The new server can now be safely started, and then any

  303.      <span class="application">rsync</span>'ed standby servers.

  304.     </p></li><li class="step"><p class="title"><strong>Post-upgrade processing</strong></p><p>

  305.      If any post-upgrade processing is required, pg_upgrade will issue

  306.      warnings as it completes. It will also generate script files that must

  307.      be run by the administrator. The script files will connect to each

  308.      database that needs post-upgrade processing. Each script should be

  309.      run using:

  310. 

  311. </p><pre class="programlisting">

  312. psql --username=postgres --file=script.sql postgres

  313. </pre><p>

  314. 

  315.      The scripts can be run in any order and can be deleted once they have

  316.      been run.

  317.     </p><div class="caution"><h3 class="title">Caution</h3><p>

  318.      In general it is unsafe to access tables referenced in rebuild scripts

  319.      until the rebuild scripts have run to completion; doing so could yield

  320.      incorrect results or poor performance. Tables not referenced in rebuild

  321.      scripts can be accessed immediately.

  322.     </p></div></li><li class="step"><p class="title"><strong>Statistics</strong></p><p>

  323.      Because optimizer statistics are not transferred by <code class="command">pg_upgrade</code>, you will

  324.      be instructed to run a command to regenerate that information at the end

  325.      of the upgrade.  You might need to set connection parameters to

  326.      match your new cluster.

  327.     </p></li><li class="step"><p class="title"><strong>Delete old cluster</strong></p><p>

  328.      Once you are satisfied with the upgrade, you can delete the old

  329.      cluster's data directories by running the script mentioned when

  330.      <code class="command">pg_upgrade</code> completes. (Automatic deletion is not

  331.      possible if you have user-defined tablespaces inside the old data

  332.      directory.)  You can also delete the old installation directories

  333.      (e.g. <code class="filename">bin</code>, <code class="filename">share</code>).

  334.     </p></li><li class="step" id="PGUPGRADE-STEP-REVERT"><p class="title"><strong>Reverting to old cluster</strong></p><p>

  335.      If, after running <code class="command">pg_upgrade</code>, you wish to revert to the old cluster,

  336.      there are several options:

  337. 

  338.      </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>

  339.         If the <code class="option">--check</code> option was used, the old cluster

  340.         was unmodified;  it can be restarted.

  341.        </p></li><li class="listitem"><p>

  342.         If the <code class="option">--link</code> option was <span class="emphasis"><em>not</em></span>

  343.         used, the old cluster was unmodified;  it can be restarted.

  344.        </p></li><li class="listitem"><p>

  345.         If the <code class="option">--link</code> option was used, the data

  346.         files might be shared between the old and new cluster:

  347. 

  348.         </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: circle; "><li class="listitem"><p>

  349.            If <code class="command">pg_upgrade</code> aborted before linking started,

  350.            the old cluster was unmodified;  it can be restarted.

  351.           </p></li><li class="listitem"><p>

  352.            If you did <span class="emphasis"><em>not</em></span> start the new cluster, the old

  353.            cluster was unmodified except that, when linking started, a

  354.            <code class="literal">.old</code> suffix was appended to

  355.            <code class="filename">$PGDATA/global/pg_control</code>.  To reuse the old

  356.            cluster, remove the <code class="filename">.old</code> suffix from

  357.            <code class="filename">$PGDATA/global/pg_control</code>; you can then restart

  358.            the old cluster.

  359.           </p></li><li class="listitem"><p>

  360.            If you did start the new cluster, it has written to shared files

  361.            and it is unsafe to use the old cluster.  The old cluster will

  362.            need to be restored from backup in this case.

  363.           </p></li></ul></div><p>

  364.        </p></li></ul></div><p>

  365.     </p></li></ol></div></div><div class="refsect1" id="id-1.9.5.12.8"><h2>Notes</h2><p>

  366.    <span class="application">pg_upgrade</span> creates various working files, such

  367.    as schema dumps, in the current working directory.  For security, be sure

  368.    that that directory is not readable or writable by any other users.

  369.   </p><p>

  370.    <span class="application">pg_upgrade</span> launches short-lived postmasters in

  371.    the old and new data directories.  Temporary Unix socket files for

  372.    communication with these postmasters are, by default, made in the current

  373.    working directory.  In some situations the path name for the current

  374.    directory might be too long to be a valid socket name.  In that case you

  375.    can use the <code class="option">-s</code> option to put the socket files in some

  376.    directory with a shorter path name.  For security, be sure that that

  377.    directory is not readable or writable by any other users.

  378.    (This is not relevant on Windows.)

  379.   </p><p>

  380.    All failure, rebuild, and reindex cases will be reported by

  381.    <span class="application">pg_upgrade</span> if they affect your installation;

  382.    post-upgrade scripts to rebuild tables and indexes will be

  383.    generated automatically.  If you are trying to automate the upgrade

  384.    of many clusters, you should find that clusters with identical database

  385.    schemas require the same post-upgrade steps for all cluster upgrades;

  386.    this is because the post-upgrade steps are based on the database

  387.    schemas, and not user data.

  388.   </p><p>

  389.    For deployment testing, create a schema-only copy of the old cluster,

  390.    insert dummy data, and upgrade that.

  391.   </p><p>

  392.    <span class="application">pg_upgrade</span> does not support upgrading of databases

  393.    containing table columns using these <code class="type">reg*</code> OID-referencing system data types:

  394.    <code class="type">regproc</code>, <code class="type">regprocedure</code>, <code class="type">regoper</code>,

  395.    <code class="type">regoperator</code>, <code class="type">regconfig</code>, and

  396.    <code class="type">regdictionary</code>.  (<code class="type">regtype</code> can be upgraded.)

  397.   </p><p>

  398.    If you are upgrading a pre-<span class="productname">PostgreSQL</span> 9.2 cluster

  399.    that uses a configuration-file-only directory, you must pass the

  400.    real data directory location to <span class="application">pg_upgrade</span>, and

  401.    pass the configuration directory location to the server, e.g.

  402.    <code class="literal">-d /real-data-directory -o '-D /configuration-directory'</code>.

  403.   </p><p>

  404.    If using a pre-9.1 old server that is using a non-default Unix-domain

  405.    socket directory or a default that differs from the default of the

  406.    new cluster, set <code class="envar">PGHOST</code> to point to the old server's socket

  407.    location.  (This is not relevant on Windows.)

  408.   </p><p>

  409.    If you want to use link mode and you do not want your old cluster

  410.    to be modified when the new cluster is started, consider using the clone mode.

  411.    If that is not available, make a copy of the

  412.    old cluster and upgrade that in link mode. To make a valid copy

  413.    of the old cluster, use <code class="command">rsync</code> to create a dirty

  414.    copy of the old cluster while the server is running, then shut down

  415.    the old server and run <code class="command">rsync --checksum</code> again to update the

  416.    copy with any changes to make it consistent.  (<code class="option">--checksum</code>

  417.    is necessary because <code class="command">rsync</code> only has file modification-time

  418.    granularity of one second.)  You might want to exclude some

  419.    files, e.g. <code class="filename">postmaster.pid</code>, as documented in <a class="xref" href="continuous-archiving.html#BACKUP-LOWLEVEL-BASE-BACKUP" title="25.3.3. Making a Base Backup Using the Low Level API">Section 25.3.3</a>.  If your file system supports

  420.    file system snapshots or copy-on-write file copies, you can use that

  421.    to make a backup of the old cluster and tablespaces, though the snapshot

  422.    and copies must be created simultaneously or while the database server

  423.    is down.

  424.   </p></div><div class="refsect1" id="id-1.9.5.12.9"><h2>See Also</h2><span class="simplelist"><a class="xref" href="app-initdb.html" title="initdb"><span class="refentrytitle">initdb</span></a>, <a class="xref" href="app-pg-ctl.html" title="pg_ctl"><span class="refentrytitle"><span class="application">pg_ctl</span></span></a>, <a class="xref" href="app-pgdump.html" title="pg_dump"><span class="refentrytitle">pg_dump</span></a>, <a class="xref" href="app-postgres.html" title="postgres"><span class="refentrytitle"><span class="application">postgres</span></span></a></span></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="pgtesttiming.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="reference-server.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="pgwaldump.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top"><span class="application">pg_test_timing</span> </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> <span class="application">pg_waldump</span></td></tr></table></div></body></html>
上海开阖软件有限公司 沪ICP备12045867号-1