summaryrefslogtreecommitdiffstats
path: root/doc/src/sgml/html/app-pgrewind.html
blob: c23f061dbaaca13863bf4bee16600b2557371c84 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!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_rewind</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 Vsnapshot" /><link rel="prev" href="app-pgresetwal.html" title="pg_resetwal" /><link rel="next" href="pgtestfsync.html" title="pg_test_fsync" /></head><body id="docContent" class="container-fluid col-10"><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center"><span class="application">pg_rewind</span></th></tr><tr><td width="10%" align="left"><a accesskey="p" href="app-pgresetwal.html" title="pg_resetwal">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 15.5 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="pgtestfsync.html" title="pg_test_fsync">Next</a></td></tr></table><hr /></div><div class="refentry" id="APP-PGREWIND"><div class="titlepage"></div><a id="id-1.9.5.9.1" class="indexterm"></a><div class="refnamediv"><h2><span class="refentrytitle"><span class="application">pg_rewind</span></span></h2><p>pg_rewind — synchronize a <span class="productname">PostgreSQL</span> data directory with another data directory that was forked from it</p></div><div class="refsynopsisdiv"><h2>Synopsis</h2><div class="cmdsynopsis"><p id="id-1.9.5.9.4.1"><code class="command">pg_rewind</code> [<em class="replaceable"><code>option</code></em>...]  { <code class="option">-D</code>  |   <code class="option">--target-pgdata</code> }<em class="replaceable"><code> directory</code></em> { <code class="option">--source-pgdata=<em class="replaceable"><code>directory</code></em></code>  |   <code class="option">--source-server=<em class="replaceable"><code>connstr</code></em></code> } </p></div></div><div class="refsect1" id="id-1.9.5.9.5"><h2>Description</h2><p>
   <span class="application">pg_rewind</span> is a tool for synchronizing a PostgreSQL cluster
   with another copy of the same cluster, after the clusters' timelines have
   diverged. A typical scenario is to bring an old primary server back online
   after failover as a standby that follows the new primary.
  </p><p>
   After a successful rewind, the state of the target data directory is
   analogous to a base backup of the source data directory. Unlike taking
   a new base backup or using a tool like <span class="application">rsync</span>,
   <span class="application">pg_rewind</span> does not require comparing or copying
   unchanged relation blocks in the cluster. Only changed blocks from existing
   relation files are copied; all other files, including new relation files,
   configuration files, and WAL segments, are copied in full. As such the
   rewind operation is significantly faster than other approaches when the
   database is large and only a small fraction of blocks differ between the
   clusters.
  </p><p>
   <span class="application">pg_rewind</span> examines the timeline histories of the source
   and target clusters to determine the point where they diverged, and
   expects to find WAL in the target cluster's <code class="filename">pg_wal</code> directory
   reaching all the way back to the point of divergence. The point of divergence
   can be found either on the target timeline, the source timeline, or their common
   ancestor. In the typical failover scenario where the target cluster was
   shut down soon after the divergence, this is not a problem, but if the
   target cluster ran for a long time after the divergence, its old WAL
   files might no longer be present. In this case, you can manually copy them
   from the WAL archive to the <code class="filename">pg_wal</code> directory, or run
   <span class="application">pg_rewind</span> with the <code class="literal">-c</code> option to
   automatically retrieve them from the WAL archive. The use of
   <span class="application">pg_rewind</span> is not limited to failover, e.g.,  a standby
   server can be promoted, run some write transactions, and then rewound
   to become a standby again.
  </p><p>
   After running <span class="application">pg_rewind</span>, WAL replay needs to
   complete for the data directory to be in a consistent state. When the
   target server is started again it will enter archive recovery and replay
   all WAL generated in the source server from the last checkpoint before
   the point of divergence. If some of the WAL was no longer available in the
   source server when <span class="application">pg_rewind</span> was run, and
   therefore could not be copied by the <span class="application">pg_rewind</span>
   session, it must be made available when the target server is started.
   This can be done by creating a <code class="filename">recovery.signal</code> file
   in the target data directory and by configuring a suitable
   <a class="xref" href="runtime-config-wal.html#GUC-RESTORE-COMMAND">restore_command</a> in
   <code class="filename">postgresql.conf</code>.
  </p><p>
   <span class="application">pg_rewind</span> requires that the target server either has
   the <a class="xref" href="runtime-config-wal.html#GUC-WAL-LOG-HINTS">wal_log_hints</a> option enabled
   in <code class="filename">postgresql.conf</code> or data checksums enabled when
   the cluster was initialized with <span class="application">initdb</span>.  Neither of these
   are currently on by default.  <a class="xref" href="runtime-config-wal.html#GUC-FULL-PAGE-WRITES">full_page_writes</a>
   must also be set to <code class="literal">on</code>, but is enabled by default.
  </p><div class="warning"><h3 class="title">Warning</h3><p>
    If <span class="application">pg_rewind</span> fails while processing, then
    the data folder of the target is likely not in a state that can be
    recovered.  In such a case, taking a new fresh backup is recommended.
   </p><p>
    As <span class="application">pg_rewind</span> copies configuration files
    entirely from the source, it may be required to correct the configuration
    used for recovery before restarting the target server, especially if
    the target is reintroduced as a standby of the source. If you restart
    the server after the rewind operation has finished but without configuring
    recovery, the target may again diverge from the primary.
   </p><p>
    <span class="application">pg_rewind</span> will fail immediately if it finds
    files it cannot write directly to.  This can happen for example when
    the source and the target server use the same file mapping for read-only
    SSL keys and certificates.  If such files are present on the target server
    it is recommended to remove them before running
    <span class="application">pg_rewind</span>.  After doing the rewind, some of
    those files may have been copied from the source, in which case it may
    be necessary to remove the data copied and restore back the set of links
    used before the rewind.
   </p></div></div><div class="refsect1" id="id-1.9.5.9.6"><h2>Options</h2><p>
    <span class="application">pg_rewind</span> accepts the following command-line
    arguments:

    </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="option">-D <em class="replaceable"><code>directory</code></em></code><br /></span><span class="term"><code class="option">--target-pgdata=<em class="replaceable"><code>directory</code></em></code></span></dt><dd><p>
        This option specifies the target data directory that is synchronized
        with the source. The target server must be shut down cleanly before
        running <span class="application">pg_rewind</span>
       </p></dd><dt><span class="term"><code class="option">--source-pgdata=<em class="replaceable"><code>directory</code></em></code></span></dt><dd><p>
        Specifies the file system path to the data directory of the source
        server to synchronize the target with. This option requires the
        source server to be cleanly shut down.
       </p></dd><dt><span class="term"><code class="option">--source-server=<em class="replaceable"><code>connstr</code></em></code></span></dt><dd><p>
        Specifies a libpq connection string to connect to the source
        <span class="productname">PostgreSQL</span> server to synchronize the target
        with. The connection must be a normal (non-replication) connection
        with a role having sufficient permissions to execute the functions
        used by <span class="application">pg_rewind</span> on the source server
        (see Notes section for details) or a superuser role.  This option
        requires the source server to be running and accepting connections.
       </p></dd><dt><span class="term"><code class="option">-R</code><br /></span><span class="term"><code class="option">--write-recovery-conf</code></span></dt><dd><p>
        Create <code class="filename">standby.signal</code> and append connection
        settings to <code class="filename">postgresql.auto.conf</code> in the output
        directory.  <code class="literal">--source-server</code> is mandatory with
        this option.
       </p></dd><dt><span class="term"><code class="option">-n</code><br /></span><span class="term"><code class="option">--dry-run</code></span></dt><dd><p>
        Do everything except actually modifying the target directory.
       </p></dd><dt><span class="term"><code class="option">-N</code><br /></span><span class="term"><code class="option">--no-sync</code></span></dt><dd><p>
        By default, <code class="command">pg_rewind</code> will wait for all files
        to be written safely to disk.  This option causes
        <code class="command">pg_rewind</code> to return without waiting, which is
        faster, but means that a subsequent operating system crash can leave
        the data directory corrupt.  Generally, this option is useful for
        testing but should not be used on a production
        installation.
       </p></dd><dt><span class="term"><code class="option">-P</code><br /></span><span class="term"><code class="option">--progress</code></span></dt><dd><p>
        Enables progress reporting. Turning this on will deliver an approximate
        progress report while copying data from the source cluster.
       </p></dd><dt><span class="term"><code class="option">-c</code><br /></span><span class="term"><code class="option">--restore-target-wal</code></span></dt><dd><p>
        Use <code class="varname">restore_command</code> defined in the target cluster
        configuration to retrieve WAL files from the WAL archive if these
        files are no longer available in the <code class="filename">pg_wal</code>
        directory.
       </p></dd><dt><span class="term"><code class="option">--config-file=<em class="replaceable"><code>filename</code></em></code></span></dt><dd><p>
        Use the specified main server configuration file for the target
        cluster. This affects <span class="application">pg_rewind</span> when
        it uses internally the <span class="application">postgres</span> command
        for the rewind operation on this cluster (when retrieving
        <code class="varname">restore_command</code> with the option
        <code class="option">-c/--restore-target-wal</code> and when forcing a
        completion of crash recovery).
       </p></dd><dt><span class="term"><code class="option">--debug</code></span></dt><dd><p>
        Print verbose debugging output that is mostly useful for developers
        debugging <span class="application">pg_rewind</span>.
       </p></dd><dt><span class="term"><code class="option">--no-ensure-shutdown</code></span></dt><dd><p>
        <span class="application">pg_rewind</span> requires that the target server
        is cleanly shut down before rewinding. By default, if the target server
        is not shut down cleanly, <span class="application">pg_rewind</span> starts
        the target server in single-user mode to complete crash recovery first,
        and stops it.
        By passing this option, <span class="application">pg_rewind</span> skips
        this and errors out immediately if the server is not cleanly shut
        down. Users are expected to handle the situation themselves in that
        case.
       </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">-?</code><br /></span><span class="term"><code class="option">--help</code></span></dt><dd><p>Show help, then exit.</p></dd></dl></div><p>
   </p></div><div class="refsect1" id="id-1.9.5.9.7"><h2>Environment</h2><p>
   When <code class="option">--source-server</code> option is used,
   <span class="application">pg_rewind</span> also uses the environment variables
   supported by <span class="application">libpq</span> (see <a class="xref" href="libpq-envars.html" title="34.15. Environment Variables">Section 34.15</a>).
  </p><p>
   The environment variable <code class="envar">PG_COLOR</code> specifies whether to use
   color in diagnostic messages. Possible values are
   <code class="literal">always</code>, <code class="literal">auto</code> and
   <code class="literal">never</code>.
  </p></div><div class="refsect1" id="id-1.9.5.9.8"><h2>Notes</h2><p>
   When executing <span class="application">pg_rewind</span> using an online
   cluster as source, a role having sufficient permissions to execute the
   functions used by <span class="application">pg_rewind</span> on the source
   cluster can be used instead of a superuser.  Here is how to create such
   a role, named <code class="literal">rewind_user</code> here:
</p><pre class="programlisting">
CREATE USER rewind_user LOGIN;
GRANT EXECUTE ON function pg_catalog.pg_ls_dir(text, boolean, boolean) TO rewind_user;
GRANT EXECUTE ON function pg_catalog.pg_stat_file(text, boolean) TO rewind_user;
GRANT EXECUTE ON function pg_catalog.pg_read_binary_file(text) TO rewind_user;
GRANT EXECUTE ON function pg_catalog.pg_read_binary_file(text, bigint, bigint, boolean) TO rewind_user;
</pre><p>
  </p><p>
   When executing <span class="application">pg_rewind</span> using an online
   cluster as source which has been recently promoted, it is necessary
   to execute a <code class="command">CHECKPOINT</code> after promotion such that its
   control file reflects up-to-date timeline information, which is used by
   <span class="application">pg_rewind</span> to check if the target cluster
   can be rewound using the designated source cluster.
  </p><div class="refsect2" id="id-1.9.5.9.8.4"><h3>How It Works</h3><p>
    The basic idea is to copy all file system-level changes from the source
    cluster to the target cluster:
   </p><div class="procedure"><ol class="procedure" type="1"><li class="step"><p>
      Scan the WAL log of the target cluster, starting from the last
      checkpoint before the point where the source cluster's timeline
      history forked off from the target cluster. For each WAL record,
      record each data block that was touched. This yields a list of all
      the data blocks that were changed in the target cluster, after the
      source cluster forked off. If some of the WAL files are no longer
      available, try re-running <span class="application">pg_rewind</span> with
      the <code class="option">-c</code> option to search for the missing files in
      the WAL archive.
     </p></li><li class="step"><p>
      Copy all those changed blocks from the source cluster to
      the target cluster, either using direct file system access
      (<code class="option">--source-pgdata</code>) or SQL (<code class="option">--source-server</code>).
      Relation files are now in a state equivalent to the moment of the last
      completed checkpoint prior to the point at which the WAL timelines of the
      source and target diverged plus the current state on the source of any
      blocks changed on the target after that divergence.
     </p></li><li class="step"><p>
      Copy all other files, including new relation files, WAL segments,
      <code class="filename">pg_xact</code>, and configuration files from the source
      cluster to the target cluster. Similarly to base backups, the contents
      of the directories <code class="filename">pg_dynshmem/</code>,
      <code class="filename">pg_notify/</code>, <code class="filename">pg_replslot/</code>,
      <code class="filename">pg_serial/</code>, <code class="filename">pg_snapshots/</code>,
      <code class="filename">pg_stat_tmp/</code>, and <code class="filename">pg_subtrans/</code>
      are omitted from the data copied from the source cluster. The files
      <code class="filename">backup_label</code>,
      <code class="filename">tablespace_map</code>,
      <code class="filename">pg_internal.init</code>,
      <code class="filename">postmaster.opts</code>, and
      <code class="filename">postmaster.pid</code>, as well as any file or directory
      beginning with <code class="filename">pgsql_tmp</code>, are omitted.
     </p></li><li class="step"><p>
      Create a <code class="filename">backup_label</code> file to begin WAL replay at
      the checkpoint created at failover and configure the
      <code class="filename">pg_control</code> file with a minimum consistency LSN
      defined as the result of <code class="literal">pg_current_wal_insert_lsn()</code>
      when rewinding from a live source or the last checkpoint LSN when
      rewinding from a stopped source.
     </p></li><li class="step"><p>
      When starting the target, <span class="productname">PostgreSQL</span> replays
      all the required WAL, resulting in a data directory in a consistent
      state.
     </p></li></ol></div></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="app-pgresetwal.html" title="pg_resetwal">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="reference-server.html" title="PostgreSQL Server Applications">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="pgtestfsync.html" title="pg_test_fsync">Next</a></td></tr><tr><td width="40%" align="left" valign="top"><span class="application">pg_resetwal</span> </td><td width="20%" align="center"><a accesskey="h" href="index.html" title="PostgreSQL 15.5 Documentation">Home</a></td><td width="40%" align="right" valign="top"> <span class="application">pg_test_fsync</span></td></tr></table></div></body></html>